Multilayered childhood adversity and mortality: a population-based cohort study of 1.2 million individuals

Summary

Background

Childhood adversity is multi-layered, extending beyond the family to include broader neighbourhood and health contexts. We aimed to investigate how these multiple layers of childhood adversity relate to the risk of death in young adulthood.

Methods

Children were followed from birth into young adulthood (16–42 years) using nationwide register data on multi-layered childhood adversity and mortality. Individual adversity included perinatal adversity (preterm or small-for-gestational-age) and mental and physical health-service use. Family adversity included five distinct groups using group-based multi-trajectory modelling based on 12 adversities. Neighbourhood adversity included material deprivation in small-area geographical zones. We evaluated associations of these layers with all-cause mortality using survival analyses.

Findings

1,235,519 individuals born between 1980 and 2001 were followed up until 31 December 2022, capturing 7320 deaths. Children facing high family adversity were more likely to have perinatal adversity, use health services, and live in deprived neighbourhoods. Each layer was separately associated with mortality; for example, high physical health-service use (HR: 2.36; 95% CI: 2.24; 2.48), corresponding to 42 (39; 45) excess deaths per 100,000 person-years (PY), while living in a deprived neighbourhood (HR: 1.20; 95% CI: 1.14; 1.26), corresponding to 8 (6; 11) excess deaths per 100,000 PY. Cross-layer additive interactions were strongest between family adversity and child health or perinatal adversity. For example, high physical health-service use combined with high family adversity were associated with 194 excess deaths (95% CI: 172–216) per 100,000 PY with an estimated 106 due to cross–layer interaction (P < 0.0001). The highest mortality risk was observed amongst those with both high family and individual adversity: HR: 7.16 (95% CI: 6.40; 8.01) compared to those with low adversity.

Interpretation

The co-occurrence and interaction of childhood adversities across different layers may create highly vulnerable groups, deepening lifelong health inequalities. This underscores the importance of a comprehensive, multi-layered approach that targets individual vulnerabilities as well as the broader social environment.

Funding

The European Research Council.

Research in context.

Evidence before this study

We searched for relevant studies in PubMed with no language restrictions from inception to 31 March 2025 using the terms (“Childhood adversity” OR “Adverse Childhood Experiences” OR “Early life adversity OR “Perinatal” OR Neighbo∗rhood deprivation”) AND (“[mortality] OR “premature mortality” OR “death” OR “suicide”). We identified studies investigating how individual forms of childhood adversity contribute to the risk of mortality in young adulthood but only a limited number of studies investigating intersections of different forms of adversity in smaller samples. We also identified several conceptual frameworks that have emphasised the importance of considering multiple, intersecting forms of adversity across developmental contexts. Empirical studies that comprehensively investigate the intersections of multiple forms of adversity simultaneously remain limited.

Added value of this study

By leveraging large-scale, multi-layer lifecourse data, this study examines emergent patterns of childhood adversity arising from co-occurring and interacting adversities across individual, family, and neighbourhood layers, and relates this to mortality in young adulthood. We demonstrate that adversity is not confined to isolated domains: children exposed to family adversity are more likely to experience perinatal complications, health difficulties, and to reside in deprived neighbourhoods. Importantly, we show that these adversities are not only independently associated with mortality but can also interact across layers, where adversity at one level may amplify the harmful effects of adversity at another. Such cross–layer interactions may help identify highly vulnerable subgroups of children who experience multiple interacting adversities already during childhood.

Implications of all the available evidence

Childhood adversity is a multi-layered phenomenon that cannot be fully understood through single-domain analyses. Our findings suggest that public-health and social policies must address adversity more broadly, considering the interplay between individual vulnerability, family adversity, and the structural conditions of the communities in which children grow up. These insights can inform more targeted interventions: from reducing premature births and supporting families in adversity to promoting mental health and tackling broader social determinants such as poverty, education, and neighbourhood conditions. By adopting a multi-layered, lifecourse approach to childhood adversity, health and social systems may better identify and support highly vulnerable subgroups of children who face adversity, ultimately reducing intergenerational cycles of disadvantage.

Introduction

Reducing health inequality remains a critical public-health priority.^1,2 The causal drivers of inequality are multifaceted and dynamic, involving multiple factors operating and interacting across different layers over the lifecourse.³ Childhood is a particularly sensitive period marked by physical, social, and cognitive development. Experiencing family-related adversity, such as poverty, disease or loss in the family during childhood, or growing up in a dysfunctional household characterised by, for example, parental mental health or substance abuse, has lifelong health consequences.4, 5, 6, 7, 8 While research on childhood adversity has largely focused on the family, some argue for a broader perspective that considers the child's wider environment.9, 10, 11, 12

The complexity of a child's social environment has long been recognised,¹³ and recent frameworks emphasise that childhood adversity can emerge across multiple layers.9, 10, 11, 12 This includes adversity within broader ecological contexts, such as neighbourhoods.^9,14 Living in a deprived neighbourhood may restrict access to adequate health-care services, healthy food environments, and safe recreational spaces. It may also induce chronic psychological stress due to a perception of threat and lack of resources, which can eventually cause physiological changes and impair health.¹⁵ Family adversity may also interact with individual vulnerabilities, such as biological susceptibility arising from being born preterm or small-for-gestational-age.¹⁶ Studies also show that some children face ongoing family adversity alongside poor mental and physical health throughout childhood.⁴ Using a complex-systems perspective, we envision the emergence of broader patterns of adversity spanning multiple layers as the result of cascading effects, in which individual health, family social conditions, and neighbourhood disadvantage may act as both causes and consequences of one another throughout childhood and across generations, ultimately contributing to lifelong disadvantage.¹⁷

Although death in young adulthood is rare and unexpected, it represents the most definitive public-health outcome. In high-income settings, leading causes of death in young adulthood include accidents and suicide followed by cancer, all of which are socially patterned and influenced by early-life adversity.⁵ This underscores the critical need to examine how childhood adversity across multiple layers may influence later mortality risk.

Testing these cross–layer interactions requires large-scale, multilayered lifecourse data. While several multi-level frameworks of childhood adversity exist,9, 10, 11, 12 they have yet to be fully operationalised and empirically validated. In this study, we aimed to investigate how multiple layers of childhood adversity relate to the risk of death in young adulthood. We hypothesised that perinatal adversity, mental-, and physical-health adversities, and neighbourhood deprivation co-occur and interact amongst individuals who experience family adversity, leading to higher mortality in young adulthood. To test these hypotheses, we used longitudinal, multi-layered data from a nationwide cohort of 1.2 million individuals followed from birth up to age 42.

Methods

The Danish Life Course (DANLIFE) cohort

The study builds on the large register-based DANLIFE cohort,¹⁸ which includes all children born in Denmark between 1980 and 2015, totalling 2,223,927 children. A unique personal identification number enables individual-level linkage across registries, allowing integration of adversity data across multiple layers. For this study, DANLIFE was expanded to include information that identifies characteristics of the neighbourhoods in which individuals grew up.¹⁹ The analysis was restricted to individuals with complete information on adversities across all layers who could be followed throughout childhood (birth years 1980–2001) and who survived to age 16 (n = 1,235,519; Figure S1).

Statistics Denmark and the Danish Health Authority granted access to pseudonymised data from the Danish administrative and health registers. The extended version of DANLIFE was approved by the Danish Data Protection Agency through the records of research projects that involve personal data at the University of Copenhagen (record no 514-0641/21-3000). The Danish Data Protection Agency ensures compliance with national and EU legislation. Registry linkage studies do not require informed consent or ethical approval by the Danish National Committee on Health Research Ethics according to Danish Law.

The multiple layers of childhood adversity

We conceptualise childhood adversity as occurring across two lifecourse dimensions (Fig. 1):

• 1.Time: Annual data from birth to age 16
• 2.Layers: Individual, family, and neighbourhood

Fig. 1.

Overview of layers and measures of childhood adversity in the study.

While it is not feasible to include every aspect of childhood adversity, we selected key indicators for each layer:

• (i)
  Individual layer. We assessed three indicators of individual adversity in childhood:

  • a.
    Perinatal adversity: Defined as being born preterm (before 37 weeks of gestation) or small-for-gestational age (below the 10th percentile on age- and sex-specific growth curves²⁰), based on the Danish Medical Birth Register.
  • b.
    Physical health: Using the National Patient Registry, we calculated total emergency and inpatient hospital contacts during childhood. High physical-health service use was defined as being in the top 20% for hospital encounters accumulated from age 0–15.
  • c.
    Mental health: Any mental-health diagnosis before age 16, as recorded in the Danish Psychiatric Central Research Register.
• (ii)Family layer. The family-adversity layer in DANLIFE has previously been established, using register linkages between children, parents, and siblings to capture a range of social adversities in five groups.⁵ An expert panel identified three key dimensions: material deprivation (family poverty, parental long-term unemployment), loss or threat of loss (serious illness in or death of a parent or sibling), and family dynamics (maternal separation, foster-care placement, parental or sibling psychiatric illness, and parental alcohol or drug abuse). Using group-based, multi-trajectory modelling of annual adversity counts across these dimensions (assuming equal weights across adversities within the same dimension) from ages 0–16, five family-adversity trajectory groups were identified: low adversity (experiencing no or few adversities), early material deprivation (experiencing poverty or parental long-term unemployment before school age), persistent material deprivation (experiencing poverty or parental long-term unemployment during entire childhood), loss or threat of loss (experiencing serious illness in or death of a parent or sibling), and high adversity (experiencing high and increasing levels of adversity across all three dimensions). Individuals were assigned to the trajectory group for which they had the highest probability of inclusion. Further details are available in the supplementary material (Table S1, Figure S2, and method description) and in Rod et al., 2020.⁵
• (iii)
  Neighbourhood layer. We built on previously generated data zones that represent small area geographic neighbourhoods of approximately 2500 inhabitants across Denmark.¹⁹ These 1885 zones, nested within 98 municipalities, were created using residential coordinates and a clustering algorithm (see Pedersen et al., 2022 for details).¹⁹ Neighbourhood material deprivation was assessed using four annual indicators (proportion of individuals age ≥25 per zone):

  • a.
    Low income (lowest gross-income quartile by year, sex, and year of birth)
  • b.
    Basic education (primary schooling only, ≤7 years if born before 1958; ≤9 years if born in 1958 and after).
  • c.
    Unemployed
  • d.
    Overcrowding (households with >1 person per room)

These measures were calculated annually from 1981 to 2014 using weighted averages from 5-year bands. Values for 1980 were set equal to those from the 1981 to 1984 band, and values for 2015–2017 were set equal to those from the 2010 to 2014 band to ensure full childhood coverage under the assumption that neighbourhood conditions change slowly year to year. For each individual, we calculated average exposure to each indicator across childhood. High exposure was defined as being in the top 20% for any indicator. High neighbourhood deprivation was defined as being in the top 20% for at least two indicators.

All-cause mortality

All individuals were followed up from age 16 until death, emigration, or end of follow-up on 31 December 2022, with mortality as the outcome of interest. Individuals who emigrated during follow-up were censored at age of emigration. Maximum follow-up was 26 years with a maximum age of 42 at the end of follow-up.

Covariates

Covariates included sex assigned at birth (male/female), birth year, maternal age at time of birth (<20 years/20–30 years/>30 years), and parental country of origin. Parental country of origin was defined as Western if one or both parents were of non-indigenous European, North American, Australian, or New Zealand descent, and non-Western if both parents were from elsewhere.

Statistical analysis

First, we characterised the patterns of adversities across the different layers. For each of the five family-adversity groups, we determined the percentage of individuals who also experienced high adversity at the individual and neighbourhood layers. We visualised these patterns as a combined radar plot and as bar plots for each specific type of adversity.

Our primary analysis consisted of three steps:

• 1.We tested associations between each layer of adversity and mortality separately. To do so, we estimated hazard ratios (HR) using Cox proportional hazard models from age 16. These relative effects enable comparison with findings from previous studies. We complemented these analyses with estimates of hazard differences (HD) from Aalen's additive hazards models to quantify the absolute magnitude of effects, which is directly relevant to public health policy. This dual reporting aligns with the STROBE statement, which notes that “expressing results in both relative and absolute terms may often be helpful”.²¹
• 2.We assessed additive cross–layer interactions between the family layer and both individual and neighbourhood layers in relation to mortality. Assessment of interaction is model dependent, and we specifically chose to use Aalen's additive hazards models because they enable detection of deviation from additivity of absolute effects. This approach allows identification of subgroups in which the largest number of excess mortality cases could potentially be prevented, making it particularly relevant for public-health and clinical decision-making.²²
• 3.We explored cumulative effects of adversities across the five family-adversity groups and according to additional layers of either individual adversity only, neighbourhood adversity only, or both. We presented this as survival plots based on the Aalen Johansen model adjusting for emigration as competing risk and 95% confidence intervals (CI) for each added layer of adversity. We also fitted both Cox hazard models and Aalen's additive models from age 16 to estimate HRs, HDs, and 95% CIs of mortality in each of these groups. The reference group was those with low family adversity and no other measured adversities.

All models were adjusted by sex, birth year, maternal age at birth, and parental country of origin with supplementary analyses stratified by sex.

Sensitivity analyses

First, we added parental education (≤10 years, 10–12 years, >12 years) as a covariate, as it is difficult to disentangle parental education and material deprivation. Second, we restricted the sample to a) individuals born from 1981 to 1998 to exclude those for whom neighbourhood data were extrapolated; and b) births from 1994 to 2001, corresponding to Denmark's transition to the International Classification of Disease (ICD)-10 system. These restricted analyses yielded similar results, supporting the robustness of our findings (results not shown). Third, while we have complete data on 96% of the sample (Table S2), we applied a weighting approach to assess the impact of excluding individuals with missing data (n = 43,018).

All data manipulation, visualisation, and statistical analyses were performed in R (version 4.4.1).²³ More information about the analyses and specific packages is available in the supplementary materials.

Role of the funding source

The funders of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the paper. The corresponding author had full access to all data in the study and had final responsibility for the decision to submit the study for publication.

Results

Co-occurrence of adversity across layers

The patterns of multilayered adversity are summarised in Fig. 2. The results are organised around the family layer. The high-adversity group has the highest proportion born small-for-gestational age, with childhood mental-health diagnoses, and with high physical health service use, whereas the persistent material deprivation group has the highest proportion living in deprived neighbourhoods (Panel A). We find a clear pattern of family adversity being associated with co-occurring adversities within all other layers (Panel B). Details about the distribution of covariates are available in Table S3.

Fig. 2.

Patterns of multi-layered adversity (Panel A) and co-occurrence of childhood adversity across multiple layers of adversity according to trajectory groups of family adversity (Panel B). Note. In panel B, each axis represents a unique indicator, and the enclosed area does not reflect a summed quantity; rather, it should be interpreted as illustrating patterns and co-occurrence across indicators.

Single-layer adversity and mortality

We recorded 7320 deaths during a mean follow-up time of 14.4 years (SD = 6.4). The leading causes of death included suicide and assault (n = 1321), accidents (n = 1291), and cancer (n = 1183). We find that each layer of adversity is associated with mortality when analysed separately (Fig. 3). For the individual layer, we show that perinatal adversity (hazard ratio (HR): 1.36; 95% CI (1.28; 1.43), corresponding to 14 (11; 17) excess deaths per 100,000 person-years (PY)), a mental-health diagnosis (HR: 3.00; 95% CI (2.82; 3.19); 68 (62; 73) excess deaths per 100,000 PY), or being a high user of physical health services (HR: 2.36.; 95% CI (2.24; 2.48); 42 (39; 45) excess deaths per 100,000 PY) during childhood are associated with a higher mortality.

Fig. 3.

Single-layer adversity and mortality. Hazard ratios (HR), hazard differences (HD) per 100,000 person-years, and 95% CI for all-cause mortality in 1,235,519 young adults (age 16–42) according to each single layer of childhood adversity. Adjusted for sex assigned at birth, maternal age at birth, parental country of origin, and birth year.

For the family layer and in line with our previous findings,⁵ we show that trajectories of family adversity are associated with higher mortality. For example, high versus low adversity within the family is associated with a HR of 3.95 (95% CI: 3.64; 4.28), corresponding to 95 (86; 104) excess deaths per 100,000 PY.

For the neighbourhood layer, we show that living in neighbourhoods with high material deprivation (HR: 1.20; 95% CI (1.14; 1.26); 8 (6; 11) excess deaths per 100,000 PY) during childhood is associated with a higher mortality risk.

Additive interactions between the family and individual layer

We find cross–layer interactions between the family and individual layer, meaning that more deaths are observed than what would have been expected by the sum of adversity at each single layer (Fig. 4, Table S4). We find that being exposed to both early perinatal adversity and high family adversity are associated with 116 excess deaths (95% CI: 98–134) per 100,000 PY compared to those with low family adversity and no perinatal adversity. Of these, 22 are estimated to be due to cross–layer interaction (P = 0.02). We also find that having both a mental-health diagnosis and high family adversity during childhood are associated with 181 excess deaths (95% CI: 157–205) per 100,000 PY compared to those with low adversity and no mental-health diagnosis. Of these, 64 are estimated to be due to interaction between adversity across layers (P < 0.0001). In addition, we find that being high users of physical-health services and experiencing high family adversity are associated with 194 excess deaths (95% CI: 172–216) per 100,000 PY compared to those with low adversity at both layers. Of these, an estimated 106 cases are due to cross–layer interaction (P < 0.0001).

Fig. 4.

Cross–layer interactions. Adjusted hazard differences (aHD) and 95% CI for all-cause mortality in 1,235,519 young adults (age 16–42) according to interactions between family adversity and perinatal adversity, high user of physical-health services, mental-health diagnosis, and neighbourhood material deprivation. Interactions are not shown when no effects are attributable to interaction. Adjusted for sex assigned at birth, maternal age at birth, parental country of origin, and birth year.

Additive interactions between the family and neighbourhood layer

We find no interaction (P = 0.53) between the family and neighbourhood layer (Fig. 4, Table S4). However, individuals exposed to both high family and neighbourhood adversity have a higher risk than the reference group with 89 excess deaths (95% CI: 72–105) per 100,000 PY due to their independent effects.

Cumulative adversity across layers and mortality

Fig. 5 shows the added mortality risk from the individual and neighbourhood layer amongst individuals in each of the five family adversity groups. The individual layer is defined as having at least one of the three indicators (perinatal adversity, high physical health service use, or mental health diagnosis). All analyses use the risk amongst those with low family adversity and no adversity across the other two layers as reference (367,002 individuals; 29.7% of sample). We find that adversity across multiple layers is associated with a higher mortality risk across all family adversity groups. For example, experiencing both individual adversity and neighbourhood adversity, despite being in the low family-adversity group, are associated with a higher mortality risk (HR: 1.82; 95% CI: 1.67; 1.98) than having low adversity across layers. The highest risk is, however, found amongst those in the high family-adversity group who also experience individual-layer adversity (HR: 7.16; 95% CI: 6.40; 8.01) compared to those with low adversity. This group includes 17,810 individuals (1.4% of the population) with 434 deaths. Table S4 shows estimates across all different combinations of adversity layers.

Fig. 5.

Cumulative adversity and mortality. Survival plots and 95% CI for all-cause mortality across the five family adversity groups and according to additional layers of either individual adversity only, neighbourhood adversity only, or both. The individual layer is defined as having one or more of the three indicators (perinatal adversity, high physical-health service use, or mental-health diagnosis). Adjusted for sex assigned at birth, maternal age at birth, parental country of origin, and birth year.

Mortality was higher in men (5126 deaths) than in women (2194 deaths). Although relative effects were similar by sex, absolute risks were greatest amongst men (see Tables S6 and S7, Figures S3 and S4). Adjusting for parental education, as well as applying weighting, only slightly reduced effect sizes, confirming the robustness of our results (Tables S8 and S9).

Discussion

Leveraging multi-layer lifecourse data, we empirically demonstrated co-occurrence of adversity across individual, family, and neighbourhood layers. Individuals experiencing high family adversity were more likely to also experience perinatal adversity, mental-health problems, be high users of physical-healthcare services, and to live in deprived neighbourhoods. We also show that each of these layers of adversity is not only individually associated with mortality in young adulthood but that some also exhibit cross–layer interactions, where adversity in one layer seems to accentuate the harmful effects of adversity in another layer. Such cross–layer interactions were found for the joint exposure to adversity at the individual (health and perinatal adversity) and the family layer. These interactions may create vulnerable population subgroups, starting in childhood and potentially constituting a strong driver of lifelong health inequality.

A child's intrafamilial environment is crucial in shaping their ability to cope and learn,⁸ and our findings align with previous research that links family-related childhood adversities to health-harming behaviours, chronic conditions, and mortality in adulthood.5, 6, 7^,24 We find that children who suffer from both individual adversity and family adversity during childhood are at a particularly high mortality risk. These findings, together with previous findings, provide empirical support for conceptualising adversity as a multi-layered construct that examines the complex interplay between the child and their environment.9, 10, 11, 12 Prior studies have recognised the multi-layered nature of adversity and tested indicators of childhood adversity across various layers.25, 26, 27 For example, a British study using data from 19,000 children identified latent profiles spanning individual, family, and environmental layers, observing that exposure to multiple forms of early adversity was associated with negative developmental outcomes in adolescence.²⁶ We extend these findings to a much larger sample, enabling direct assessment of cross–scale interactions. Additionally, we utilise data spanning over four decades to track mortality into adult life.

Our findings underscore the importance of neighbourhood factors on health. Although neighbourhood deprivation was less strongly associated with mortality than adversity at other layers, it still contributed independently. Children spend most of their time in small local areas, and there is growing interest in how living in certain neighbourhoods can impact health and contribute to health inequalities.¹⁵ A recent study found that individuals of low socioeconomic status living in deprived neighbourhoods faced a higher risk of self-harm and violent crime compared to those in more affluent areas.²⁵ We demonstrate how neighbourhood material deprivation co-occurs with family adversity and is independently associated with mortality. These results highlight the need for policies aimed at addressing neighbourhood deprivation in addition to other types of adversity. Such policies could include interventions targeting housing, living conditions, poverty, and crime prevention.²⁸ Future research should explore the dynamic nature of adversity; for example, by examining how individual and family adversity changes due to residential mobility. This could offer insight into how adversity unfolds and when interventions may be most effective.

The observed excess all-cause mortality may reflect heterogeneous underlying pathways, some of which differ across causes of death, while others may be shared (for example toxic stress responses, limited contextual resources). These heterogeneous pathways have been suggested to manifest in variation in susceptibility and capacity to respond to adversity, depending on history of exposure.²⁹ While family adversity shows the strongest association with mortality, our findings indicate that all layers of adversity present potential risk, even amongst children who are not exposed to family adversity. Preventive strategies should extend beyond addressing only individuals at high risk and also target vulnerable subgroups experiencing disadvantages in any adversity layer. This approach requires expanding services from an individual focus to delivering integrated support for entire families who experience complex needs.³⁰ Policies should prioritise building contextual resources—such as social support, institutional responsiveness, and access to services—to strengthen the ability of children, families, and neighbourhoods with adversity to adapt and recover, as well as supporting multi-level interventions that increase health equity by expanding options, not just reducing exposures.¹¹

The multi-layered data infrastructure covering almost complete national lifecourse data over four decades in more than 1.2 million individuals provided unique opportunities for exploring interactions between adversities across layers. Further, the study benefitted from previous Danish work on small-area level neighbourhoods for a detailed characterisation of neighbourhoods.²⁷ We have focused on neighbourhood material deprivation in the current study, but future studies could expand this to also include measures of, for example, social fragmentation and marginalisation.

We operationalised adversity based on selected key indicators. These indicators were carefully selected based on the social-ecological model and the bio-exposome model,^9,13 but we were limited by the availability of data in the registers. For example, information on undiagnosed health conditions, housing conditions, or neighbourhood physical environment were not available through Danish registries. While the most severe cases of child abuse and neglect were likely captured through foster-care records, we lacked information on unreported instances occurring within families. We were also unable to include adversity measures related to broader ecological contexts, such as air pollution, racism, bullying, and economic crises. While we believe that our approach allows for a comprehensive examination of childhood adversity across relevant layers, we acknowledge that we may not have captured all important dimensions of childhood adversity, thus leading to an underestimation of effects.

We focused on all-cause mortality as the most definitive public-health outcome in this young population. Utilising the linked nationwide data largely mitigated selection bias. However, we excluded children who died before age 16, some of whose deaths were likely linked to perinatal adversity. Although child mortality in Denmark is very low (<1%), this exclusion may have influenced the results, which should therefore be interpreted as conditional on survival to age 16. The most common causes of death contributing to the observed mortality pattern were suicide, accidents, and cancer. While previous studies have documented associations between childhood adversity and suicide, fewer have examined accident- and cancer-related mortality in this context. Future research should explore in greater detail the underlying mechanisms, disease trajectories, and cause-specific mortality patterns—including how factors across layers cause and reinforce each other—to better understand the overall mortality pattern identified in the present study.

Denmark's strong social-security system may limit the generalisability of these findings to other socio-political contexts. Thus, our results likely represent a ‘lower bound scenario’ within a setting with government-subsidised, universal health care. These findings highlight the childhood origins of health inequality and emphasise the need for upstream structural interventions to prevent adversity and its health consequences across different countries.

Conclusion

Addressing childhood adversity requires a comprehensive, multi-layered approach. We show that multiple adversities across individual, family, and neighbourhood layers co-occur, which may contribute to persistent disadvantage. Our findings highlight the need for health-inequality research and interventions that address the full spectrum of adversity. Overlooking these interconnected factors carries the risk of oversimplifying complex health issues and potentially limiting intervention effectiveness. By recognising that childhood adversity goes beyond the family context, we can integrate policies across multiple levels—including policies to reduce premature births, support disadvantaged families, promote mental health, and address broader social determinants like poverty, education, and neighbourhood characteristics—which may help improve outcomes for children and families in vulnerable positions.

Contributors

NHR, SKB, LKE, AGZ, and TRV conceptualised the study and its design. DK and TRV did data linkage and data cleaning. SKB, LKE, AGZ, and TRV performed the analyses for the study. NHR, SKB, LKE, DK, AGZ, and TRV had access to all study data. NHR wrote the first and final draft of the manuscript. All authors discussed the results and implications of the study. TRV, SKB, DK, LKE, AGZ and NHR had access to and verified the underlying data. All authors reviewed the manuscript, and have seen and approved the final manuscript.

Data sharing statement

Individual participant data used in this study is not publicly available. Enquiries about secure access to the DANLIFE data under conditions stipulated by the Danish Data Protection Agency can be directed to principal investigator of the study, Naja Hulvej Rod (nahuro@sund.ku.dk). An interactive map of the data zones can be found online (ncrr-au.shinyapps.io/data_zones/).

Declaration of generative AI and AI-assisted technologies in the manuscript preparation process

During the preparation of this work, the authors used perplexity.ai in order to improve readability and language. After using this service, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Declaration of interests

NHR acknowledges funding from the Lundbeck Foundation (grant no. R396-2022-352). LKE acknowledges funding from the Independent Research Fund Denmark (grant no. 10.46540/4310-00032B). TRV received a travel grant from the Life History Research Society and received support for travel from the ROCKWOOL Foundation (Grant number 1255). DTR acknowledges funding from Wellcome Trust, the Medical Research Council (MRC), UK CF Trust, and the Health Foundation. DTR alsothat his team received paid commission fees for an expert report prepared and presented to UK COVID Inquiry through the University of Liverpool. This report is unrelated to the current work.