From One Death to Another: The Relationship between Familial Deaths and One’s Own Mortality Risk Among an Urban Black American Cohort

Abstract

Black Americans continue to be 1.5 times more likely to experience premature death with life expectancy up to six years shorter than their white American counterparts. These racial disparities in mortality translate into Black Americans being much more likely to experience the deaths of family members at younger ages in the life course. This study examines the impact of experiencing familial death on the survivor’s mortality risk among a cohort of Black men and women. Data collected from a community cohort first assessed in 1966 (at age 6) and followed at three additional time points (ages 16, 32, and 42) are supplemented with mortality data, retrieved from the National Death Index, that include deaths through 2021 (modal age 61). Among the 941 participants who survived to age 32 and had information on familial deaths, 38.9% experienced the death of a parent, child, or sibling by age 32, and close to one-fifth (18.2%) died between ages 33 and 61. Cox regression models that adjust for early life covariates revealed a 48% higher mortality risk among those who experienced at least one familial death by age 32; separate models provide evidence that the accumulation of familial deaths is related to midlife mortality risk. Models of relationship type indicate that death of a mother or sibling is associated with a 74% and 77% increase in mortality risk, respectively. Results highlight the heavy burden of premature familial mortality on Black Americans and its adverse impact on one’s own life expectancy.

INTRODUCTION

Despite consistent increases in life expectancy for Black Americans prior to COVID-19, the Black-White gap in life expectancy remained close to four years (Arias and Xu, 2020; Schwandt et al., 2021; Shiels et al., 2017), resulting in an estimated 1.63 million excess deaths among Black Americans from 1999-2020 (Caraballo et al., 2023). Whereas post-pandemic estimates indicate an overall reduction in life expectancy for all Americans, the life expectancy disparity is now closer to six years shorter for Black Americans than white Americans (70.8 vs. 76.4; Andrasfay and Goldman, 2021, 2022; Arias et al., 2023). These racial disparities in mortality reflect not only Black Americans’ own increased risk of premature mortality but also their disproportionate exposure to the premature death of family and household members at comparatively younger ages than white Americans. Indeed, evidence from nationally representative data sets show that Black Americans are more likely to experience the death of a household member (Dixon, 2024), more likely to experience the loss of a close family member (i.e., parent, spouse, sibling, child), and more likely to experience multiple familial deaths than white Americans (Umberson et al., 2017).

The principle of linked lives from the life course perspective states that individuals are embedded in interdependent systems of social relationships and the formulation and dissolution of these social connections shape one’s life course (Carr, 2018; Elder, 1985). Whereas the research base is replete with how the formulation of social connections promote health, well-being, and longevity (Carr et al., 2014; Kiecolt-Glaser and Newton, 2001; Koball et al., 2010; Thomas et al., 2017; Holt-Lunstad and Smith, 2012), comparatively less studied is how the loss of social relationships, such as through the death of a loved one, impacts one’s own premature mortality.

Loss of a loved one can lead to the loss of social support, increased risk of loneliness, stress, grief, and neglect of own’s health (Umberson, 2017), which in turn diminish one’s health and ultimate longevity. Indeed, evidence spanning the past several decades has repeatedly found that divorce and the loss of a spouse can increase risk of death (Liu et al., 2020; Sbarra et al., 2011; Stroebe et al., 2007) and that the death of a parent, child, and sibling may increase mortality risk in the survivor (Donnelly et al, 2020a; Debiasi et al., 2021; Espinosa and Evans, 2013; Li et al., 2003; Rostila et al., 2012; Song et al., 2019), yet the research base on how the disproportionate exposure to loss among Black men and women may impact their own mortality has only recently begun to emerge (e.g., Donnelly et al, 2020a; Liu et al, 2020).

Bereavement and Mortality

Against a backdrop of genetic and environmental risks that are shared by families and households, several psychological and social factors link familial/household death and one’s own mortality risk (Stroebe et al., 2007). One primary pathway is through stress proliferation (Pearlin et al., 2005), which can lead to physiobiological changes such as increased inflammation and compromised immunity (O’Connor et al., 2021). Stress in the wake of familial loss can result from a loss of financial support, which can initiate a cascade of financial strains, school non-completion (Patterson et al., 2020; Thyden et al., 2020), and exacerbated socioeconomic disadvantage—all fundamental causes of health risks (Link and Phelan, 1995). Loss of a loved one can also lead to engagement in poor coping mechanisms, such as overeating, smoking, alcohol, or substance use (Umberson et al., 2008), which relate to increased mortality risk (Mokdad et al., 2004).

These consequences of familial death may be particularly deleterious for Black young adults who tend to experience more “complicated” grief in the wake of familial loss than white Americans (Laurie and Neimeyer, 2008) due to an acute awareness of structural racism and society’s unfair contribution to the loss, resulting in anger. Also, the loss is often accompanied by a propensity to grieve in private and show strength in public rather than seek professional or other support (Laurie and Neimeyer, 2008; Rosenblatt and Wallace, 2005).

The bereavement experience of Black Americans may contribute to ‘weathering,’ which Geronimus (2001) describes as early health deterioration resulting from an accumulation of adverse social and economic experiences, such as repeated financial strain, excessive family and kin obligations, and disruption of the family unit (e.g., early death, divorce). These multiple stressors engender unhealthy coping behaviors, high effort coping, and biological alterations that progressively erode one’s health over the life course and potentially hasten mortality (Geronimus, 2001; Geronimus et al., 2006). Although the process of weathering can apply to anyone experiencing an accumulation of adverse experiences, Geronimus posits that this process is particularly salient for Black women, suggesting a differential impact by sex.

Drawing on these mechanisms of bereavment, several recent studies using the national Health and Retirement Study data find that the disproportionate familial loss experiences among Black Americans are associated with multiple health risks, such as cardiometabolic health (Donnelly et al., 2022), low subjective life expectancy (Donnelly et al., 2020b), and dementia risk (Cha et al., 2022). These findings further suggest that the unequal distribution of these experiences among Black Americans is a potential driver of racial health disparities. What has been less studied is how this disproportionate exposure to familial loss among Black Americans relates to the ultimate outcome of one’s own mortality risk as well (for exceptions, see Donnelly et al, 2020a; Liu et al, 2020).

Life Course Variability in Familial Death Experiences

The life course perspective (Elder, 1985), research on stressful life events and bereavement (Dohrenwend et al., 1990; Institute of Medicine, 1984; Thoits, 2010; Wheaton, 1990), and literature on the effects of familial loss on health outcomes (e.g., Donnelly et al., 2022; Liu et al., 2022) direct us to examine the potential sources of heterogeneity in the impact of experiencing familial death on one’s own mortality among Black men and women. First, shorter life expectancies of Black Americans increase exposure to multiple losses earlier in the life course, which can influence the grieving experience (Rosenblatt and Wallace, 2005). Similar to the notion that risk factors tend to cluster within the same individual (Rutter, 1979), the idea of cumulative disadvantage posits that stressful life events, including the death of a loved one, tend to cluster in individuals, perhaps through genetic influences, shared environmental exposures, or unique experiences, which results in deleterious consequences (Dannefer, 1987, 2003; Ross and Wu, 1996). Moreover, Jackson and colleagues (2010) find that despite evidence of early resiliency among Black Americans, the accumulation of adverse experiences (e.g., poverty, discrimination) tend to take their toll in adulthood. Thus, the repeated exposure to social connection loss from multiple deaths over the life course might be more detrimental than experiencing fewer or more isolated events (Lewis et al., 2021).

Another potential source of variation is the timing of the loss as outcomes may depend on when in one’s life course an event occurs (Braveman and Barclay, 2009; Elder, 1985). Experiencing parental death during childhood may be particularly deleterious as it disrupts the childhood home and prematurely severs attachments to caregivers (Kamis et al., 2022). Although the supportive networks and multigenerational households common among Black families may increase one’s exposure to familial or household deaths (Dixon, 2024), these same household structures and availability of social supports may offset the impact of the disruption caused by a familial loss (Mutran, 1985; Nobles, 2007), particularly that of a parent or sibling. Alternatively, experiencing parental death in young adulthood may be particularly distressing as young adults often rely on their parents for financial and emotional support as they navigate the transition to adulthood, and this type of support is more difficult to replace upon the death of a parent in adulthood.

Variation may also stem from the type of relationship lost. A robust research base finds that familial loss from a wide variety of relationship types has adverse physical and mental effects (e.g., Liu et al., 2020; Marks et al., 2007; Rostila et al., 2012; Song et al., 2019); yet, these studies tend to focus on health rather than the survivor’s mortality risk and those that do focus on mortality risk (e.g., Donnelly et al., 2020a; Liu et al, 2020) have focused on one type of familial loss rather than a host of relationship types simultaneously. Moreover, the existing literature’s primary focus on bereavement from spousal loss (Stroebe et al., 2007) also limits our understanding of the influence of other types of close family member loss among Black Americans, who are less likely to marry (Mayol-García et al., 2021). Overall, gaining a better understanding of the influence of the exposure, accumulation, timing, and type of familial death on mortality risk among Black Americans could greatly improve strategies for culturally-tailored interventions and, in turn, mortality outcomes.

Current Study

In the present study, we ask two interrelated research questions. First, is exposure to familial death earlier in the life course associated with the risk of premature mortality in midlife? Second, does this relationship differ based on the accumulation, timing, and/or type of familial loss? In examining these questions, the current study makes several contributions. First, this study investigates the relationship between familial loss and mortality, rather more proximal health conditions. Second, we use longitudinal data from a community cohort of Black Americans who were assessed at four time points across the life course, ranging from ages 6 to 42. In the United States, it is difficult to disentangle socioeconomic position and race/ethnicity as these are highly correlated due to historical and current systemic racism practices. Focusing on one community cohort who were raised in the same neighborhood ostensibly “controls” for race, age, and early neighborhood exposures that could influence mortality risk, allowing us to better isolate the influence of familial loss.

Third, whereas examinations comparing health and mortality between racial groups are instrumental in documenting health disparities, examining the extent of familial loss and its association with premature mortality within a community cohort of Black men and women advances our understanding of how these experiences may contribute to the mortality burden of Black Americans (see Taylor et al., 2021a). Fourth, the life course and stress perspectives direct us to the notion of variability in how individuals are affected by similar life events. In this study, we examine multiple dimensions of familial death experiences, including whether the accumulation and timing of exposure to familial death over the life course are particularly detrimental to one’s mortality risk, as well as whether the type of relationship is a source of variability in mortality risk.

Finally, the life course perspective emphasizes the importance of early life experiences in shaping life’s trajectories (Elder, 1985; Hayward and Gorman, 2004). Although we cannot tease out whether the relationships are due to shared genetics/environments or causal processes, the prospective, longitudinal nature of the data allows us to better isolate the associations under study by controlling for several shared genetic and environmental risks.

DATA AND METHODS

Woodlawn Study

The Woodlawn Study follows a cohort of 1,242 Black Americans who attended first grade in 1966-67 in Woodlawn, a neighborhood in Chicago, Illinois (see Doherty and Green, 2023). In the 1960s, as a result of white flight out of city centers, the great migration of Black Americans to the north, and racist real estate practices (e.g., redlining), Woodlawn was a predominantly Black American community (98%; Kellam et al., 1975). Woodlawn was one of the most socially disadvantaged communities in Chicago at the time (e.g., 21% of the residents received public assistance, which was three times higher than for the city of Chicago), although the neighborhood was also economically heterogeneous; 47% lived above the poverty level, and 42% of the mothers had completed twelve or more years of education.

During the 1966-67 school year, all first graders at the 12 schools in Woodlawn were recruited to participate in the study (N=1,242). Only 13 families declined. That year, teachers reported on the child’s adaptations to school, and mothers or mother surrogates reported on the family’s social and economic resources, childrearing practices, and psychological and behavioral aspects of the child (Kellam et al., 1975). In adolescence, the mothers (n=939) and adolescents (n=705) living in the Chicago area provided extensive information on the family, behaviors, and psychological well-being (Ensminger and Slusarcick, 1992). The original cohort was again interviewed at ages 32 (n= 952) and 42 (n=833) and were asked about their families, community involvement, drug and criminal behavior, and physical and mental health. The historical context from first grade to adulthood for this cohort is particularly relevant to their mortality risk. For instance, the HIV/AIDS and crack epidemics began in the 1980s, when the cohort members were in their 20s, and rates of gun violence and mass incarceration sharply increased during the 1990s, when the cohort members were in their 30s. These contexts of increased mortality risk were countered by the introduction of welfare reform during their 30s and the Affordable Care Act during their 40s, both of which enabled families to meet their basic needs and broadened access to health care.

For this study, data are drawn from the age 6 and adolescent interviews to assess early parental loss (birth to age 16) and early life covariates, as well as from the young adult interview to assess deaths of immediate family members who were living in the adolescent household (ages 17 to 32). Our analytic sample is limited to the 941 (52.5% women and 47.5% men) who completed the household roster portion of the age 32 interview. Those who had left home by age 16 did not complete the household roster. This analytic sample, which represents 78.8% of the 1,194 who survived to age 32, is likely a healthier, more educated, and less transient group than the full cohort as it excludes those who died young (n=49), and includes those who are more likely to be high school graduates (57% vs. 31%) and have fewer moves as a child (2.82 vs. 2.16) than those who were living but not interviewed in young adulthood.

Measures

Mortality:

One’s own mortality is defined as death after age 32 using data from the National Death Index (NDI), which include deaths through 2021. Cohort members were coded as deceased with an accompanying age of death (if they died after age 32 and before the end of 2021) or coded as alive if they did not die by the end of 2021. Although three participants were aged 62 when they died in 2021, for ease of presentation we refer to midlife mortality as ages 33 to 61 throughout the manuscript as the majority of cohort members were 61 in 2021. Close to one-fifth of the analytic sample died between ages 33 and 61 (18.2%, n=171; 15.4% of women and 21.3% of men).

We focus on midlife mortality for two reasons. First, we use data from the age 32 interview to measure key independent variables of familial loss (described below), which restricts the sample to those who survived until the age 32 interview. Second, focusing on mortality from ages 33 to 61 allows for the inclusion of both deaths from external causes (e.g., overdose, homicide), which capture a behavioral pathway, and internal causes (e.g., cardiovascular problems), resulting perhaps from stress and bereavement, such as inflammation and compromised immune systems.

Data on Familial Death:

Data on familial deaths, defined as mother/father, sibling, or child deaths, are gathered from multiple assessments across the life course. Mother and father deaths occurring before age 17, which are drawn from the age 6 and adolescent interviews, include deaths of a biological parent living in the childhood household. At ages 6 (1966) and 16 (1976), the mother or mother surrogate reported whether the reason for a “natural” mother or father not being in the household was because they were deceased. The parental deaths captured in adolescence are restricted to parents who had been living in the childhood household. For those who lost a parent before age 17, the majority had been living with the deceased parent at age 6 (96% of those who lost a mother and 68% of those who lost a father). Sensitivity analyses inclusive of residential and non-residential parental deaths are consistent with those presented. Sibling deaths occurring in childhood and adolescence were not systematically captured during these assessments and thus could not be captured.

Deaths of one’s mother/father (inclusive of step, adoptive, and foster) or sibling (inclusive of step and half) who were living in the adolescent household are drawn from the young adulthood interview (1992) where cohort members recounted the members of their adolescent household and whether they were alive or dead at the time of the interview.

Separately, at the young adult interview, participants were asked if any of their children had died and, if so, how many. Although participants reported on spousal death as well, these are excluded as only 5 cohort members reported the loss of a spouse by age 32, which may be attributed in part to low marriage rates in this cohort.

Exposure to Any Familial Death

Exposure is defined as experiencing any familial death between birth and age 32. This includes death of a 1) biological mother or father living in the childhood household, 2) biological, step, or adoptive/foster mother or father living in the adolescent household, 3) sibling living in the adolescent household, or 4) child.

Accumulation of Familial Death

Accumulation is measured as the number of familial deaths from birth to age 32 and ranges from 0 to 6. It is determined by summing: 1) the number of parental losses in childhood (i.e., biological mothers/fathers living in the childhood household) and parental losses in young adulthood (e.g., biological, step, adoptive/foster who lived in the adolescent household), which ranged from 0 to 2; it should be noted that the vast majority of parental losses were of biological parents (95.4% of mothers who died and 75.4% of fathers); 2) the number of sibling losses, which ranged from 0 to 3; and 3) the number of child losses by 32, which ranged from 0 to 6.

Timing of Parental Death

We explored the timing of parental deaths (during childhood/adolescnce versus young adulthood) using two three-category variables: Timing of Mother’s Death and Timing of Father’s Death, coded as no loss (=0), loss in childhood/adolescence (=1), or loss in young adulthood (=2). Whereas no one lost a mother in both childhood and young adulthood, five cohort members lost a father figure in both developmental periods. These five participants are coded “1” to prioritize the first paternal death.

Type of Familial Loss

To measure type of familial loss, we created four dichotomous variables: Death of a Mother from birth to age 32, Death of a Father from birth to age 32, Death of a Sibling from ages 17 to 32, and Death of a Child by age 32. Although respondents could experience multiple losses of each relationship type, we chose to dichotomize these variables for comparability; 89.2% of those who lost a sibling lost one, and 80.0% of those who lost a child lost one. Eighty-nine percent of the analytic sample were raised with a sibling in the household, and 74% had a child. Those without a sibling or child were coded as “0” under the assumption that they did not experience that type of familial loss.

Early Life Covariates:

We include several covariates drawn from the first grade and adolescent assessments that tap into the structural circumstances, shared life experiences, and genetic characteristics that can shape one’s mortality risk across the life course. First, we include the sex of the respondent (female=0, male=1). Second, to assess early health, we include low birthweight, which is measured using 10 categories ranging from 0=8 pounds or more to 9=less than 3.5 pounds, and whether or not the participant had a chronic health condition in childhood as reported by the mother or mother surrogate. Third, because some genetically-rooted chronic conditions may manifest later in life, we include a measure of whether the participant’s mother or father had a chronic condition in 1966. Fourth, we include a measure of early life poverty that captures whether the household income was at or below the U.S. government poverty threshold based on family size in first grade and/or in adolescence. Fifth, we include high school non-completion (0=high school graduate or GED; 1=high school non-completion), based on self-reports in young adulthood and Board of Education administrative data. Sixth, early life residential instability is captured by the number of residential moves between birth and 1975 as reported by the mother or mother surrogate. Although residential instability could be a result of early familial loss that influences later mortality, we operationalize it as a covariate in this study given the frequency of residential moves among this cohort and the fact that the majority of familial losses occurred during young adulthood. Seventh, to tap into risks stemming from shared environments, we include a measure of whether an adult in the household was a regular smoker during the participant’s adolescence. Finally, we include the household size in adolescence (not including oneself), which ranges from 1 to 14, to control for the fact that exposure to death may be related to kin size.

Analytic Strategy

We conduct a series of descriptive analyses to estimate the prevalence of one’s own mortality as well as the dimensions of familial loss among this cohort followed by bivariate analyses (i.e., X² and t-tests) for the total sample and by sex to examine the risk of mortality for those who do and do not experience familial loss across the various dimensions. We also use nonparametric survival curves to estimate the cumulative risk of mortality for those who do and do not experience familial loss. Finally, we conduct a series of Cox proportional hazard models to accommodate the censored data inherent in examining mortality in a multivariable framework and account for the multiple early life covariates that could influence both exposure to familial death and one’s own premature mortality. In separate models, we regress mortality risk on the exposure, accumulation, timing, and type of family death, controlling for early life course covariates for the full analytic sample. We report adjusted hazard ratios and p-values. We present the timing of parental death results using “0” as the reference category to compare experiencing a parental loss in each time period to no loss in the tables and report supplemental results that compare experiencing a parental loss in young adulthood to a loss in childhood or adolescence in the text. Although there are no missing data on key variables and the majority of the covariates have minimal missingness, covariates drawn from the adolescent interview have up to 20% missing. Therefore, we use 40 multiply imputed datasets within the statistical package STATA (StataCorp, 2021) when developing estimates to maximize power and decrease bias (Graham et al., 2007).

RESULTS

Extent of Mortality and Familial Loss

To begin to answer the question of whether various dimensions of exposure to familial loss in early life are associated with the risk of mortality in midlife, we first describe the extent of mortality and the extent of familial loss within the cohort. As shown in Table 1, 18.2% of the analytic sample died between the ages of 33 and 61 (n=171), with men more likely to die than women (21.3% and 15.4%, respectively). The mean age of death was 50.04 (sd=7.92) with similar ages for men and women (49.73 and 50.44, respectively). The majority of the mortality was from physical illness, such as cancer or cardiovascular disease (79.3% of the 164 whose cause of death is known), as opposed to an external cause (e.g., homicide, overdose, suicide; 20.7%). Comparisons by sex show that men were more likely to die from external causes than women in midlife (26.7% and 13.5%, respectively).

Table 1:

Descriptive Statistics on Key Study Variables, Total Sample and By Sex

		Total (n=941)	Men (n=447)	Women (n=494)	Test statistic, p-value
Mortality	Percent of Cohort Members Who Died (ages 33 to 61)	18.2%	21.3%	15.4%	X2=5.43, p=.020
	Mean (sd) Age of Death	50.04 (7.92)	49.73 (7.94)	50.44 (7.92)	t=0.59, p=.556
Familial Lossa	Percent Experiencing Any Familial Death (birth to age 32)	38.9%	35.8%	41.7%	X2=3.44, p=.063
	Mean (sd) Number of Familial Deaths (birth to age 32; range 0-6)	0.53 (0.78)	0.48 (0.74)	0.57 (0.81)	t=1.81, p=.071
	Percent Experiencing Death of Mother (birth to age 32)	14.1%	13.0%	15.2%	X2=0.94, p=.332
	Death of Mother from birth to age 16	2.7%	2.2%	3.0%	X2=1.10, p=.576
	Death of Mother from ages 17 to 32	11.5%	10.7%	12.2%
	Percent Experiencing Death of Father (birth to age 32)	21.6%	21.3%	21.9%	X2=0.05, p=.820
	Death of Father from birth to age 16	8.3%	8.5%	8.1%	X2=0.24, p=.887
	Death of Father from ages 17 to 32	13.8%	12.8%	13.8%
	Percent Experiencing Death of Sibling (ages 17 to 32)	6.9%	7.8%	6.1%	X2=1.13, p=.288
	Percent Experiencing Death of Child (by age 32)	6.9%	4.5%	9.1%	X2=7.84, p=.005
Early Life Covariatesb	Mean (sd) Low Birthweight (range=0 to 9)	2.43 (2.32)	2.19 (2.26)	2.64 (2.35)	t=−2.94, p=.003
	Percent with Chronic Condition	3.6%	4.8%	2.5%	t=1.89, p=.059
	Percent with Parent Chronic Condition	13.1%	13.2%	12.9%	t=0.15, p=.878
	Percent with Early Life Poverty	62.6%	64.3%	61.0%	t=1.12, p=.263
	Percent High School Non-Completion	19.9%	22.4%	17.7%	t=1.91, p=.056
	Mean (sd) Residential Moves (range=0 to 13)	3.43 (2.32)	3.50 (2.31)	3.37 (2.33)	t=0.50, p=.620
	Percent with Regular Smoker in Adolescent Household	53.7%	56.6%	51.0%	t=1.50, p=.134
	Mean (sd) Adol. Household Size (range=1 to 14)	4.84 (2.25)	4.73 (2.26)	4.94 (2.23)	t=−1.45, p=.148

^aFrom birth to age 16, Mother includes biological mother living in the childhood household; Father includes biological father living in the childhood household. From ages 17 to 32, Mother includes mother, stepmother, adoptive mother, or foster mother living in the adolescent household; Father includes father, stepfather, adoptive father, or foster father living in the adolescent household. Sibling includes sister, brother, step-sister or brother, and half-sister or brother living in the adolescent household.

^bDue to the use of multiply imputed data for the early life covariates, test statistics and p-values are based on bivariate regression using gender as a predictor of each covariate.

Over one-third of the analytic sample experienced the loss of at least one family member by age 32 (38.9%). Death of a parent was common (31.0%) and is more than 10 percentage points higher than comparable national estimates (Weaver, 2019), revealing the cohort’s disproportionate exposure to familial death. Table 1 also shows that exposure to familial loss accumulated with a mean of 0.53 losses by age 32; 27.8% of the sample experienced one familial death, 8.7% experienced two, and 2.3% experienced three or more by the age of 32. Women in the cohort were more likely to experience a familial loss and had more familial deaths than the men, but the associations were not statistically significant.

As for type, paternal deaths were most common with 21.6% of the sample losing a father by age 32, followed by loss of a mother (14.1%). Exposure to parental deaths was more likely during young adulthood with 11.5% of the sample losing a mother and 13.8% losing a father between the ages of 17 and 32, compared with 2.7% and 8.3% losing a mother or father, respectively, in childhood or adolescence. Seven percent of the sample lost a sibling, and 7% lost a child, with women more likely to report losing a child than men.

We next examine the bivariate relationship between exposure to familial loss in early life with one’s own mortality risk in midlife, shown in Table 2. Among those experiencing a familial death, more than one in five (22.7%) died by 61, compared with 15.3% of those who did not. Kaplan-Meier survival curves estimating the time to death show that mortality risk between those who did and did not experience a familial loss begins to diverge in the mid-40s before widening further in the mid-50s (see Figure 1).

Table 2:

A Comparison of the Probability of Dying Between Ages 33 and 61 by Familial Death Experiences (n=941)

	Prevalence of Death: Ages 33 to 61
	Among Those Who Experienced Familial Death	Among Those Who Did Not Experience Familial Death	Test Statistic, p-value
Any Familial Death (birth to age 32)	22.7%	15.3%	X2=8.177, p=.004
Death of Mother (birth to age 32)	29.3%	16.3%	X2=12.953, p<.001
Death of Mother (birth to age 16)	24.0%	18.0%	X2=.587, p=.444
Death of Mother (ages 17 to 32)	30.6%	16.6%	X2=12.582, p<.001
Death of Father (birth to age 32)	17.7%	18.3%	X2=.003, p=.855
Death of Father (birth to age 16)	15.4%	18.4%	X2=.444, p=.505
Death of Father (ages 17 to 32)	19.2%	18.0%	X2=.102, p=.749
Death of Sibling (ages 17 to 32)	27.7%	17.5%	X2=4.256, p=.039
Death of Child by age 32	26.2%	17.6%	X2=2.991, p=.084

Figure 1: Timing of One’s Own Death (Ages 33 to 61*) Stratified by Experiencing Any Familial Death (Birth to Age 32).

*Three of the participants who died were age 62 at the time of death.

Chi-square tests of independence between mortality risk and type of familial death across the life course reveal that these differences appear to be primarily driven by experiencing the loss of a mother or sibling in young adulthood. As shown in Table 2, close to one-third (30.6%) of those experiencing a maternal death between ages 17 and 32 died, compared with 16.6% who did not experience a maternal death during this time. Experiencing the death of a sibling who lived in the adolescent household was also significantly associated with mortality risk compared with those who did not experience this type of loss (27.7% and 17.5%, respectively). Whereas estimates for experiencing the death of a child are similar to those for death of a sibling, these are not statistically significantly different at p<.05.

Relationship between Exposure to Familial Loss and Mortality Risk

To begin to answer the core research question of whether various dimensions of exposure to familial loss are associated with the risk of mortality in midlife, we estimate a series of Cox regression models that control for early life covariates. Table 3 shows that those who experienced a familial death between birth and age 32 had close to 1.5 times the mortality risk of those who did not (Model 1, aHR=1.481, p=.012). This estimate translates to a 48% higher mortality risk for those experiencing at least one familial death by young adulthood.

Table 3:

Multiply Imputed Cox Regression Models: Dimensions of Familial Death and One’s Own Mortality Risk (n=941)

	Model 1: Exposure		Model 2: Accumulation		Model 3: Timing		Model 4: Relationship Type
	Adjusted Hazard Ratio	p-value	Adjusted Hazard Ratio	p-value	Adjusted Hazard Ratio	p-value	Adjusted Hazard Ratio	p-value
Familial Death Dimensions
Any Familial Death (birth to age 32)	1.481	.012	---	---	---	---	---	---
Number of Familial Deaths (birth to age 32)	---	---	1.298	.003	---	---	---	---
Timing of Death of Mother
Mother Alive at age 32	---	---	---	---	ref.	ref.	---	---
Death of Mother (birth to ages 16)	---	---	---	---	1.251	.599	---	---
Death of Mother (ages 17 to 32)	---	---	---	---	1.894	.001	---	---
Timing of Death of Father
Father Alive at age 32	---	---	---	---	ref.	ref.	---	---
Death of Father (birth to age 16)	---	---	---	---	0.690	.225	---	---
Death of Father (ages 17 to 32)	---	---	---	---	1.060	.797	---	---
Death of Mother (birth to age 32)	---	---	---	---	---	---	1.738	.003
Death of Father (birth to age 32)	---	---	---	---	---	---	0.884	.520
Death of Sibling (ages 17 to 32)	---	---	---	---	---	---	1.773	.027
Death of Child (by age 32)	---	---	---	---	---	---	1.375	.229
Covariates
Male	1.427	.023	1.441	.020	1.395	.033	1.425	.027
Low Birthweight	1.052	.101	1.053	.097	1.042	.189	1.049	.123
Chronic Childhood Condition	1.487	.252	1.459	.275	1.394	.340	1.389	.344
Parent with Chronic Health Condition	1.095	.682	1.082	.723	1.101	.668	1.088	.707
Early Poverty	1.399	.057	1.399	.056	1.400	.058	1.358	.085
High School Non-Completion	1.802	.001	1.769	.001	1.845	.001	1.724	.002
Number of Residential Moves	0.949	.144	0.947	.125	0.946	.126	0.946	.120
Regular Smoker in Adolescent Household	0.855	.366	0.852	.355	0.862	.391	0.861	.389
Adolescent Household Size	0.964	.295	0.963	.286	0.971	.405	0.960	.254

Moreover, the hazard of midlife mortality was significantly associated with the number of familial deaths (Model 2, aHR=1.298, p=.003), with a 30% higher risk of premature mortality for each additional familial death experienced. To examine whether this effect is reflecting exposure to any familial loss as opposed to accumulation, we re-ran the multiply imputed models for Model 2 using a categorical variable of experiencing no deaths, one death, or 2 or more deaths. The results (not shown) are consistent with the idea of accumulation, as evidenced through a positive and marginally significant hazard ratio comparing one death to none (aHR=1.401, p=.053) and a further elevated positive and significant hazard ratio when comparing two or more deaths to none (aHR=1.691, p=.019). However, the results also point to exposure to the first loss as a primary driver, as evidenced by a lower and non-significant hazard ratio when comparing two or more deaths to one death (aHR=1,207, p=.424).

With respect to timing, Model 3 estimates the effect of the loss of a household mother or father on one’s own premature mortality. No significant association between paternal loss and one’s own mortality was found, regardless of timing, and the effect of maternal loss in childhood/adolescence was non-significant. However, the hazard of midlife mortality was significantly higher for those experiencing a maternal death in young adulthood compared with those whose mother was alive at age 32 (aHR=1.894, p=.001). A comparison between losing a mother in young adulthood and losing a mother in childhood or adolescence shows a higher but non-significant difference in mortality (aHR=1.514, p=.357, results not shown). Model 4 shows that experiencing a mother’s death and a sibling’s death is related to nearly twice the midlife mortality risk when compared to those who did not experience these types of deaths (aHR= 1.738, p=.003 and aHR= 1.713, p=.027, respectively), with non-significant effects for paternal or child deaths.

Finally, given the significant sex differences in the prevalence of one’s own mortality, we would be remiss if we did not examine sex differences in the relationship between familial loss and mortality risk. A series of regression models that included an interaction term between each familial death dimension and sex showed no statistically significant interactions, indicating that the presented conclusions do not differ by sex (results available upon request).

DISCUSSION

By explicating how premature mortality “begets” premature mortality among a community cohort of Black men and women, this study begins to unpack how exposure, accumulation, timing, and type of familial loss through young adulthood predicts one’s own mortality in midlife. Several findings provide key contributions to the literature. First, this study confirms that exposure to familial loss, defined as experiencing the death of a parent, sibling, or child by age 32, is more common among an urban cohort of Black Americans than national estimates, although most did not experience a familial loss by age 32 (61%). Second, results show that experiencing multiple familial losses is associated with midlife mortality risk, adding to the burgeoning literature finding that multiple family deaths are deleterious to one’s health (e.g., Donnelly et al., 2022; Liu et al., 2022).

Third, our findings indicate that the type of relationship lost has distinct effects, with the loss of a mother or sibling significantly increasing midlife mortality risk. Notably, the loss of a mother appears to have age-graded effects on midlife mortality, as experiencing the death of a mother during one’s young adult life, but not during childhood or adolescence, was significantly associated with higher mortality risk. Although this finding is counter to an abundance of literature that notes the detrimental long-term impact of experiencing loss in childhood and adolescence, particularly parental loss (e.g., Debiasi et al., 2021), it aligns with literature citing the role of social support as a key buffer to stressful life event experiences (Thoits, 2010), such as the loss of a parent, and the more limited literature pointing to young adulthood as a sensitive period for exposure to loss (Kamis et al., 2022). A potential explanation of this finding is grounded in research on the availability and supportive nature of both kin and non-kin networks and multigenerational households commonly found in Black families and communities (Mutran, 1985; Nobles, 2007). These strong networks within communities that disproportionately experience loss may provide particularly salient buffers of the developmental damage associated with the loss of a mother in childhood and adolescence. Moreover, experiencing familial loss in young adulthood may be especially detrimental if the individual relies on that family member for direct support (e.g., receiving childcare support) or if that individual is expected to provide support for younger family members (e.g., provide caregiving support for siblings).

The loss of one’s mother as particularly damaging may stem from the stable fact that for the past several decades, Black men have been dying approximately 7 years younger than Black women (Arias, 2011; Arias et al., 2023), resulting in a greater likelihood that a mother’s death is the second parental loss, thus producing exponential effects (Horowitz et al., 1984). In this cohort, whereas the majority still had both parents alive at age 32, 26% had lost one parent and 5% had lost two parents. Although our data cannot isolate the temporal ordering of these dual parental losses within each life stage, prevalence estimates across the life stages show that only one of the 25 people who lost their mother before age 16 lost their father in young adulthood (4%), yet 18% of those who lost a father by age 16 lost their mother in young adulthood. Another explanation is rooted in the fact that families often rely on childcare and other instrumental support from grandmothers (i.e., the primary caregiver’s mother) during young adulthood, particularly families living in poverty and where the mother is employed (Baydar and Brooks-Gunn, 1998; Laughlin, 2013), resulting in a stronger potential for a cascade of hardships and weathering (Geronimus, 2001) in the wake of the loss of one’s mother than one’s father.

The emergence of sibling loss as a predictor of midlife mortality risk highlights the potential importance of this relationship during the transition to adulthood (Gilligan et al., 2020). This finding builds upon the existing evidence based on registry data from Sweden and Denmark that finds experiencing the death of a sibling during childhood increases risk of mortality (Rostila et al., 2017; Yu et al., 2017) and cardiovascular disease (Huang et al., 2024). This study extends the timeline of experiencing sibling death well into young adulthood and expands the generalizability to include Black men and women. Given that living with a sibling is very common (i.e., 77% of children under 18 in the US lived with a sibling under the age of 25 in 1995; Hernandez, 1997), coupled with attachment theories that posit the strength and value of the sibling relationship in adulthood (van Volkom, 2006), more research on the impact of this type of loss on one’s own mortality is needed.

We were not able to fully account for genetic and shared experiences that may increase the risk of both familial death and one’s own death, precluding us from making causal conclusions. There is an acute need for further research that can not only isolate the causal influence but also can examine the mechanisms that connect familial loss to mortality among Black Americans, as well as possible moderators. Unfortunately, data limitations did not allow us to examine the degree to which the loss affected economic, emotional, and social trajectories, which would help identify such mechanisms. We were also unable to explore potential variability in mortality risk by cause or lag between familial death and one’s own death, which would provide insight on causal mechanisms. Future research could explore the role of positive coping behaviors (e.g., spiritual coping, increased community engagement) and increased awareness of one’s health (e.g., early screening behaviors) in the wake of a familial death, especially as they relate to various causes of familial death (e.g., hereditary causes, suicide, or homicide). Finally, the data limited us from examining losses from siblings prior to age 17, non-immediate family household members, close non-household relatives (e.g., extended family), or close friends throughout the life course, which represent important exclusions in light of the evidence of increased heterogeneity in family/household structures (Dixon, 2024; Bengtson, 2001; Kellam et al., 1977) and the importance of fictive kin relationships among Black Americans (Chatters et al., 1994; Taylor et al., 2021b).

Despite limitations, this study advances racial health disparities research by exposing the broad reach of mortality disparities and examining how experiencing familial death relates to one’s own mortality risk. As this study’s intent is to help advance our understanding of familial death’s toll on the mortality burden among Black Americans, we were able to examine potential variation in the effects across the life course dimensions of exposure, accumulation, timing of familial loss, and type of relationship loss. The results support previous research that excessive experiences of familial loss contribute, in part, to the perpetuation of racial disparities in mortality. Importantly, these findings highlight that interventions to address disparities in premature mortality may have potential compounding effects by not only saving one life but multiple lives through the prevention of grief and potential loss of life among loved ones.

HIGHLIGHTS.

• Exposure to familial death is prevalent among this cohort.

• Exposure and accumulation of familial loss is related to midlife mortality risk.

• Maternal or sibling loss is significantly associated with one’s own mortality risk.

• Young adulthood may be a particularly sensitive period for the impact of loss.