The Longitudinal Associations Between Paternal Incarceration and Family Well-Being: Implications for Ethnic-Racial Disparities in Health

Abstract

Objective:

Ethnic-racial minority children in the United States are more likely to experience father loss to incarceration than White children, and limited research has examined the health implications of these ethnic-racial disparities. Telomere length is a biomarker of chronic stress that is predictive of adverse health outcomes. We examined whether paternal incarceration predicted telomere length shortening among youth from childhood to adolescence, whether maternal depression mediated the link, and whether ethnicity-race moderated our results.

Method:

Research participants were 2,395 families in the Fragile Families and Child Well-being Study, a national and longitudinal cohort study of primarily low-income families from 20 large cities in the United States. Key constructs were measured when children were on average ages 9 (2007–2010) and 15 (2014–2017).

Results:

Children who experienced paternal incarceration exhibited shorter telomere lengths between ages 9 and 15, and changes in maternal depression mediated our finding. Specifically, mothers who experienced a partner’s incarceration were more likely to have depression between children’s ages 9 and 15. In turn, increases in maternal depression between children’s ages 9 and 15 predicted more accelerated telomere length shortening among children during this period. Paternal incarceration was more prevalent and frequent for ethnic-racial minority youth than for White youth.

Conclusion:

Paternal incarceration is associated with a biomarker of chronic stress among children in low-income families. Rates of paternal incarceration were more prevalent and frequent among Black American and multi-ethnic-racial families than among White Americans. As a result, the criminal justice system’s mass incarceration crisis is likely shaping intergenerational ethnic-racial health disparities.

Introduction

Deeply rooted racism in the United States has produced ethnic-racial inequality in the criminal justice system,¹ which disproportionately impacts ethnic-racial minority children’s developmental trajectories.² In the U.S., 25% of Black children—as compared to only 4% of White children—have experienced father loss to the criminal justice system (i.e., paternal incarceration).³ Considering the adversity that father loss can cause to developing children, unjust practices contributing to these disparities must be considered as a threat to these children’s future social capital and public health.²

In response to these growing concerns, the present study contributes to the literature on racial inequality and paternal incarceration. First, most studies have combined paternal incarceration with other adverse childhood experiences (ACEs; e.g., maternal depression, parental substance abuse, and other forms of father loss).^{4, 5} We argue that these ACEs may represent distinct but related dimensions of children’s ecologies; as such, we distinguish paternal incarceration in isolation from other ACEs to understand how it specifically shapes families’ well-being. Second, childhood represents a period when individuals are most sensitive to toxic stressors, which can increase their susceptibility to disease during adulthood.⁶ For instance, children with incarcerated fathers are likely to receive psychiatric diagnoses and experience health problems during adulthood.^{7, 8} We examine children’s telomere lengths (TLs) as possible biological mechanisms linking paternal incarceration during childhood to the onset of health morbidities in adulthood.^{9, 10}

Third, the process of how paternal incarceration affects children through parenting has been theorized,⁶ but few empirical studies have captured this process. Considering that parenting is modifiable to intervention, we assess whether maternal depression mediates the link between paternal incarceration and youth’s telomeres to guide interventions. Finally, only a handful of studies have examined ethnic-racial variation in the impacts of paternal incarceration on youth’s adjustment.^{11, 12} We build on extant studies and consider how racism within the criminal justice system can have implications for disparities in the next generation of children’s health.

Paternal Incarceration and Children’s Telomere Lengths

Paternal incarceration exposes children to the trauma of father loss,¹³ economic instability,¹⁴ disruptive parenting,¹² and parental separation.¹⁵ Families may also experience logistical and financial barriers to jail/prison visits; in addition, when families are able to make such visits, jail/prison settings are developmentally inappropriate for children.^{16, 17} During jail/prison visits, children undergo personal searches, abide to no-contact policies, and endure harsh treatment by jail/prison staff,¹⁷ all of which discourage future visits.^{13, 17} After release, fathers’ challenges with the labor market strain family relationships.¹⁸ Children must also navigate the stigma associated with paternal incarceration and contend with challenges of living in an economically disadvantaged household.^{13, 19} In turn, these socio-cultural processes may result in children keeping paternal incarceration a secret from their school peers and educators, which can stoke feelings of ostracism.¹⁹ Thus, paternal incarceration is a toxic stressor for children that may contribute to psychological and behavioral maladjustment.^{20, 21}

Within the ecobiodevelopmental (EBD) framework, toxic stress during childhood has consequences for physical health and well-being across the lifespan.⁶ Toxic stressors impair brain architecture and burden the hypothalamic-pituitary-adrenocortical axis and the sympathetic-adrenomedullary system.²² In turn, chronic physiological stress responses heighten children’s allostatic load that can be evidenced by shortened TLs.⁶ To capture children’s embodiment of stress, we examine their TLs as possible mechanisms through which paternal incarceration shapes children’s developmental life courses.

Telomeres are repetitive DNA sequences that serve as protective caps on the ends of chromosomes, confer protection to underlying genetic material, and prevent chromosomes from fusing with one another.²³ Telomeres progressively shorten with each cell division during DNA replication and are indicators of cellular aging. TL has been used to quantify the biological effects of environmental stress during childhood, including father loss.^{24, 25} Shortened telomeres compromise genetic sensitivity within the cell and promote auxiliary cellular processes that further decrease TL with age and stress.²³ Long-term repercussions of shortened telomeres include heart disease,¹⁰ cancer,⁹ diabetes,²⁶ and greater all-cause mortality rates.²⁷ The possible role of TLs in the relation between paternal incarceration and health morbidities can inform interventions looking to reduce youth’s toxic stressors associated with paternal incarceration.

Maternal Depression as a Mediator

We use the EBD model⁶ to theorize that paternal incarceration affects children’s TLs via maternal depression. Mothers who lose their partners to incarceration often experience uncertainty regarding the length of their partners’ imprisonment, rehabilitation, and societal reintegration.¹³ Mothers also experience changes in their romantic relationships and have less time for children due to role strain, which can affect their mental health.¹³ Although mothers’ psychopathology may not predict alterations in child-mother relationships, poor maternal mental health may influence parenting, including increased engagement in punitive discipline and decreased parental supervision.¹³

Extant studies suggest that the relations among paternal incarceration, maternal depression, and youth’s TLs are plausible. Mothers report a greater risk for depression due to the economic and parenting challenges following their partners’ incarceration.²⁸ Moreover, children growing up with depressed mothers have demonstrated more accelerated TL shortening,^{29, 30} and scholars attribute this phenomenon to the stressful context in the home and children’s exposure to maternal cognition, behaviors, and affect.³¹

Ethnic-Racial Group Differences

Ethnic-racial minority men are disproportionately represented in the American legal system.^{2, 18} Due to the interaction between socioeconomic status and ethnicity-race in the U.S., not only are ethnic-racial minority men more likely than White men to be taken into custody, but they are less able to pay bail and are more likely to be incarcerated as pre-trial detainees.¹⁸ Ethnic-racial minority defendants are more likely to receive lengthier sentences than White defendants.¹⁸

In turn, these ethnic-racial disparities in paternal incarceration shape the trajectory of family functioning. Lengthier incarceration sentences and discrimination in the labor market have predicted greater parental separation for Black families than for White families.¹⁵ These factors associated with incarceration lower an ethnic-racial minority couple’s perception of marriage as a viable option and limit mothers’ opportunities to find other marriageable men.³² Even when mothers re-partner with their returning, formerly incarcerated partners, these families are more likely to reside in communities with limited supportive resources (i.e., poverty, substance abuse).³² After a father is released, a family’s income has dropped by 15%.² Hence, social and economic stressors prior to and following paternal incarceration can shape worse effects for ethnic-racial minority households than for White households.

The Present Study

Considering paternal incarceration yields stress and uncertainty, we hypothesized that youth who experienced paternal incarceration would demonstrate accelerated TL shortening. However, considering paternal incarceration often impacts mothers’ emotional and conventional support, we expected that maternal depression would mediate the link between paternal incarceration and children’s TLs. Specifically, when a father is incarcerated, we postulated that mothers would report a greater risk for depression, which in turn would predict accelerated TL shortening among children over time. Thus, we tested the following research questions: (1) Do children who experience paternal incarceration exhibit accelerated TL shortening from childhood to adolescence; (2) does maternal depression mediate the link between paternal incarceration and children’s TL shortening; and (3) does ethnicity-race moderate the rates and risks of paternal incarceration? In sensitivity analyses, we tested whether our results differed between families with residential and nonresidential fathers and examined the direction of effects among our key constructs.

Method

Our data included 2,395 Black, White, and multi-ethnic-racial families. These families participated in the Fragile Families and Child Well-being (FFCW) study,³³ a longitudinal and national cohort study of 4,898 children born in the U.S. with primarily unmarried parents (exclusionary criteria are available upon request). The FFCW study’s strengths include comprehensive measures of key constructs among families in socially disadvantaged communities.

Table 1 presents demographic characteristics across ethnic-racial groups. No ethnic-racial group differences emerged for children’s sex, χ²(2)=.21, p=.90, paternal death, χ²(2)=1.92, p=.38, or maternal incarceration, χ²(2)=3.32, p=.19. White and multi-ethnic-racial families were more likely to experience a parental separation/divorce relative to Black families, χ²(2)=12.61, p<.01. Black and White youth were younger than multi-ethnic-racial youth, but Black and White youth did not differ from each other on age, F(2, 2392)=7.94, p<.001. White mothers had more advanced degrees than multi-ethnic-racial mothers, who had more advanced degrees than Black mothers, F(2, 2392)=63.69, p<.001. Fathers in multi-ethnic-racial households were less likely to be U.S. citizens than their Black and White peers, χ²(2)=37.11, p<.001. More White parents were married, χ²(2)=473.82, p<.001, whereas more Black and multi-ethnic-racial parents were co-habitants, χ²(2)=14.55, p<.001. More White and multi-ethnic-racial fathers struggled with alcohol use, χ²(2)=102.14, p<.001, but more Black fathers struggled with drug use, χ²(2)=18.04, p<.001. Black and White families lived in census tracts with more ethnically-racially congruent residents than multi-ethnic-racial families [Black residents: F(2, 2392)=663.08, p<.001; White residents: F(2, 2392)=736.09, p<.001]. Black families lived in census tracts with more adults without high school degrees than multi-ethnic-racial families, who lived in census tracts with more adults without high school degrees than White families, F(2, 2392)=169.44, p<.001. Overall, families lived in ethnically-racially congruent neighborhoods, and more White families lived in more socioeconomically affluent neighborhoods than Black and multi-ethnic-racial families.

Table 1:

Demographic Information for the Full Sample and by Families’ Ethnicity-Race

Variables’ mean-levels	All families (n = 2395)	Black American families (n = 1448)	White American families (n = 494)	Multi-ethnic-racial families (n = 453)
Youth age	9.27 (.39)	9.26 (.40)a	9.22 (.32)a	9.33 (.46)b
%Female youth	48.00	48.60	46.80	48.10
%Maternal education
Less than high school	15.00	18.00a	9.10b	17.00a, b
High school	21.00	23.50a	18.20b	21.00a, b
Some college	18.00	50.50a	33.60b	42.60a
College or beyond	46.00	8.10a	39.10b	19.40c
%Other adverse childhood experiences
Mother incarcerated	8.20	8.50	6.30	9.30
Father alcohol use	25.20	17.40a	45.50b	27.80b
Father drug use	12.80	13.90a	9.50b	12.60a
%Parental/Paternal status
Baseline parents married	23.20	11.50a	59.30b	21.00c
Baseline parents cohabitated	33.40	34.60a	26.50b	37.30a
Paternal U.S. Citizen	93.80	95.50a	95.10a	87.00b
Father separated/divorced	10.90	9.00a	13.20b	14.10b
Father deceased	1.30	1.60	0.80	1.10
Age-1 father lifetime incarceration	4.30	5.60a	1.00b	4.00a, b
%Neighborhood-level characteristics
White census tract residents	42.83	24.90a	77.12b	46.22c
Black census tract residents	39.71	59.85a	8.53b	25.87c
Adult residents without high school degrees	19.89	22.82a	12.04b	19.08c
School-level characteristics
#Out-of-school suspensions	3.19	20.00a	7.35b	12.92c
#Reported race-based bullying	0.27	0.17a	0.30a, b	0.57b
#Full-time teachers	39.00	35.00a	51.00b	38.00a, b
#School size	540.00	515.00a	586.00b	573.00b
%Black students	49.05	65.53a	14.88b	31.73c
%White students	28.54	16.15a	64.69b	32.21c
Key study variables (% or mean, SD)
Paternal incarceration age 9	8.20	10.30a	3.80b	5.30c
Paternal incarceration age 15	17.80	20.30a	10.30b	17.90c
Maternal depression age 9	17.60	17.00	17.50	18.80
Maternal depression age 15	17.80	17.40	18.40	19.40
Youth TL (ln) age 9	2.03 (.32)	2.05 (.31)	2.01 (.34)	2.02 (.33)
Youth TL (ln) age 15	2.08 (.28)	2.08 (.29)	2.08 (.27)	2.08 (.27)

Note: When time is unspecified, variables were assessed at the age-9 assessment. Ln = natural logarithm; TL = telomere length.

^aMeans differ from values with subscripts^b and^c at p < .05.

^bMeans differ from values with subscripts^a and^c at p < .05.

^cMeans differ from values with subscripts^a and^b at p < .05.

Procedure

The FFCW study randomly sampled 4,898 births from 75 hospitals in 20 large cities between 1998 and 2000 with an oversample of nonmarital births (at a ratio of 3:1). Mothers and fathers were interviewed soon after the birth of the focal child, and follow-up interviews were conducted when children were approximately ages 1 (1999–2001), 3 (2001–2003), 5 (2003–2006), 9 (2007–2010), and 15 (2014–2017). The present study used two waves of data when children were on average ages 9 and 15, which were the only waves when saliva samples were collected to assess children’s TLs. No ethics approval was sought, as the University of Pittsburgh’s Institutional Review Board does not require review for the analysis of de-identified, publicly available data [exempt Criteria 45-CFR-46.104(d)(4)].

As part of the ages 9 and 15 assessments, the FFCW study’s in-home module included the collection of genetic data in addition to the survey collected from children. Salivary DNA samples were taken using the Oragene DNA sample collection kit (DNA Genotek Inc, Ottawa, ON). In cases where a biological father or non-parental figure was the primary caregiver or the biological mother was not present for the in-home visit, a saliva sample was collected from the child only. Families completing the home visit activities received a $65 payment to the primary caregiver and a $30 payment to the child. Ultimately, 86% of children who participated in the age-9 in-home survey provided saliva samples for genotyping.

Measures

Paternal incarceration.

Survey contractors’ disposition information and parental surveys were combined to assess paternal incarceration.³⁴ The age-9 assessment represented whether paternal incarceration occurred at age 9, and the age-15 assessment represented whether paternal incarceration occurred between ages 9 and 15 (0=no, 1=yes).

Telomere measurement.

TLs at ages 9 and 15 were determined using a qPCR method that incorporates a double-stranded oligomer standard to permit the measurement of absolute TL in kilobases per chromosome^{24, 35} (see Supplement 1, available online, for details regarding TL production). Outliers were dealt with by trimming 1% off both tails of the sample; after we evaluated TL at each wave for normalcy, we estimated the natural log transformation for TL at each wave to correct for the positive skew of the data.^{24, 35}

Maternal depression.

The Composite International Diagnostic Interview–Short Form³⁶ was used to assess maternal major depressive episodes in the past year at children’s ages 9 and 15 assessments. Maternal depression represented whether mothers (0=no, 1=yes) had depression and/or anhedonia feelings in the past year that lasted at least two weeks; if so, mothers reported whether the symptoms (i.e., feeling sad, blue, or depressed, losing interest, feeling tired, changing weight, trouble sleeping, trouble concentrating, feeling worthless, and thinking about death) lasted at least half the day. As a conservative test of our hypotheses, this binary variable omits mothers with minimal or mild depressive symptoms³⁷ and who could have contributed to the link between paternal incarceration and youth’s telomeres.

Covariates.

We accounted for several third variable confounds that could bias the link between paternal incarceration and youth’s TLs. Unless specified otherwise, all covariates were assessed at age 9 and included maternal education, paternal U.S. citizen status, other confounding ACEs (i.e., maternal incarceration, paternal alcohol and drug use), baseline parental marital/cohabitation status, other forms of father loss (i.e., separation/divorce, death, and age-1 lifetime incarceration), neighborhood factors (i.e., census data on the percentages of Black and White residents and the percentage of residents ages 25 and up without high school degrees), and school-level characteristics (i.e., number of out-of-school suspensions, number of reported race-based bullying/harassment allegations, number of full-time equivalent teachers, school size, and the percentages of Black and White students). Youth’s covariates included their age, sex, body mass index, pubertal development, smoking, problematic sleep duration, depressive symptoms, self-rated health, school suspensions/expulsions, grade retention, and parental-school involvement (see Supplement 2, available online, for measurement details).

Missing Data

Among the analytic sample (n=2,395), only 118 youth had one wave of data missing, which was at age 15. No significant relationships emerged between study participation and families’ ethnicity-race, maternal education, children’s sex, neighborhood ethnic-racial and socioeconomic composition, or key constructs (i.e., paternal incarceration, maternal depression, and youth’s telomeres; detailed comparisons are available upon request). To retain sample variability and diversity, we used full information maximum likelihood to handle missing data, which performs better than other missing data approaches when missing data patterns are random.³⁸

Analytic Plan

All analyses were conducted in Mplus version 8.3³⁹ with TYPE=COMPLEX to account for families nested in cities and using maximum likelihood with robust standard errors to account for the non-normal distribution of the data. To answer our research questions, we used multi-level models that assigned time to Level 1 and families to Level 2 while accounting for city random effects. The intra-class correlations in Table 2 justified our multi-level framework, as most of the variance for each construct was at the within-and between-family levels relative to the between-city level. To answer our first research question, we examined whether changes in paternal incarceration predicted changes in youth’s log-transformed TLs when maternal depression was not included in the equation. To answer our second research question, we explored the relations between change scores for paternal incarceration, maternal depression, and youth’s log-transformed TLs. In doing so, we tested whether changes in paternal incarceration predicted changes in youth’s log-transformed TLs via changes in maternal depression between ages 9 and 15 by using a non-parametric bootstrapping technique to generate standard errors for the indirect effect via the multivariate delta method.⁴⁰ This quasi-experimental framework allowed us to treat each research participant like his/her own control group and examine the degree to which families experienced deviations relative to their own average on an outcome when a father was incarcerated.

Table 2:

Intra-Class Correlations Indicating the Percent of Variance at the Within-Family, Between-Family, and Between-City Levels for All Key Study Constructs

Measure	Within Family	Between Family	Between City
ΔPaternal incarceration between age 9 to age 15	76	23	1
ΔYouth TL (ln) between age 9 to age 15	50	49	1
ΔMaternal depression between age 9 to age 15	70	29	1

Note: Ln = natural logarithm; TL = telomere length.

Results

Table 1 includes descriptive statistics of key constructs, and Table 3 includes zero-order bivariate correlations among them (see Table S1, available online, for correlations between covariates and key constructs). Eight percent of children experienced paternal incarceration at age 9, while 18% did at age 15. In addition, 6% of the sample experienced paternal incarceration at both waves, 15% experienced paternal incarceration at only one wave, and 79% did not experience paternal incarceration at either wave. Between ages 9 and 15, log transformed TLs declined for 46% of youth and increased for 54% of youth (see Figure 1).

Table 3:

Zero-Order Bivariate Correlations among Key Study Variables

	Variables	1	2	3	4	5	6
1	Paternal incarceration age 9	1
2	Paternal incarceration age 15	.42**	1
3	Maternal depression age 9	.04*	.10**	1
4	Maternal depression age 15	.05*	.09**	.29**	1
5	Youth TL (ln) age 9	−.05*	−.01	−.08**	−.04	1
6	Youth TL (ln) age 15	−.04	−.01	−.03	−.05	.50**	1

Note: Ln = natural logarithm; TL = telomere length.

^*p < .05;

^**p < .01.

Figure 1:

Spaghetti Plots Depicting Each Child’s Change in the Log Transformation of Telomere Length between Ages 9 and 15

Note: Kb = kilobases per chromosome; Ln = natural logarithm.

Table 1 also presents key constructs across ethnic-racial groups. Black and multi-ethnic-racial families were more likely to experience paternal incarceration than White families at each wave [age-9: Δχ²(1)= 26.68, p<.001; age-15: Δχ²(1)=41.41, p<.001]. Black families were also more likely to experience paternal incarceration than multi-ethnic-racial families at each wave [age-9: Δχ²(1)=10.36, p<.01; age-15: Δχ²(1)=4.45, p<.05]. Although we found no mean-level differences in log transformed TLs among ethnic-racial groups at each wave [age-9: F(2, 1691)=2.35, p=.10; age-15: F(2, 1560)=.20, p=.82], we estimated a difference score to compare change scores in log transformed TLs over time among our ethnic-racial groups. This difference score showed that White youth were more likely to experience increases in TLs than their peers (r=.06, p<.05). No reliable ethnic-racial group differences emerged for maternal depression at each wave [age-9: Δχ²(2)=.54, p=.77; age-15: Δχ²(2)=1.08, p=.58].

Paternal Incarceration and Youth’s Telomere Lengths

In Figure 2, after controlling for covariates and city random effects, children who experienced paternal incarceration showed diminished log-transformed TLs over time (B=−.04, SE=.02, 95%CI [−.11, −.01], t=−2.07, p=.03).

Figure 2:

A Multi-Level Mediation Analysis Examining the Mediating Role of Maternal Depression between Paternal Incarceration and Youth Telomere Lengths between Ages 9 and 15

Note: Ln = natural logarithm.

*p < .05; **p < .01.

Maternal Depression as a Mediator

Figure 2 presents estimates for the inter-relations among change scores for paternal incarceration, maternal depression, and youth’s log-transformed TLs between ages 9 and 15 after we controlled for covariates and city random effects. Mothers who lost their partners to incarceration were more susceptible to depression between ages 9 and 15 (B=.07, SE=.02, 95%CI [.03, .10], t=3.64, p<.001). In turn, increases in maternal depression between ages 9 and 15 predicted accelerated log-transformed TL shortening among youth during this period (B=−.04, SE=.01, 95%CI [−.06, −.02], t=−3.39, p=.001). The indirect effect of paternal incarceration on shortened log-transformed TLs via changes in maternal depression was significant (B=−.01, SE=.00, 95%CI [−.01, −.01], t=−3.16, p=.002). After standardizing the observed significant effects to produce effect sizes, paternal incarceration had small effects on children’s log-transformed TLs (.06) and maternal depression (.07). According to the effect size for TL, paternal incarceration was associated with a 6% reduction in TL between ages 9 and 15. The model fit the data well, χ²(216)=767.85, p<.001, RMSEA=.02, CFI=.93, SRMR_within=.00, SRMR_between=.03.

Ethnic-Racial Group Differences

To examine whether our results were stronger for specific ethnic-racial groups, we shifted families’ ethnicity-race as a covariate in Figure 2 to a grouping variable for multi-group analyses. After we stratified our results by families’ ethnicity-race (see Table S2, available online), we found that constraining the observed effects to be equivalent across groups did not result in a significant decrement in model fit when examining maternal depression [Δχ²(2)=1.22, p=.54] and children’s TLs [Δχ²(2)=5.88, p=.09] as outcomes, suggesting that ethnicity-race did not moderate our results.

However, while paternal incarceration’s effects may not be disproportionately large among particular ethnic-racial groups, this finding says nothing about concentrations or exposure rates. Ethnic-racial disparities in paternal incarceration rates may concentrate deleterious effects among particular ethnic-racial groups if they are disproportionately exposed to the stressor. Indeed, Black and multi-ethnic-racial families were more likely than their White peers to experience paternal incarceration at each wave. Black families were also more likely than their multi-ethnic-racial peers to experience such loss at each wave. Specifically, 24% of Black and 20% of multi-ethnic-racial families with incarcerated fathers experienced the observed negative links between paternal incarceration and families’ well-being, whereas only 12% of White families did, χ²(2)=32.74, p<.001. In addition, 7% of Black and 4% of multi-ethnic-racial families experienced paternal incarceration at least twice, whereas only 2% of White families did, χ²(2)=39.06, p<.001, indicating that these ethnic-racial minority families experienced paternal incarceration more frequently than White families.

Sensitivity Analyses

We examined whether baseline paternal marital/cohabitation status moderated the effects of paternal incarceration on maternal depression and youth’s TLs. Paternal incarceration predicted greater maternal depression and shorter youth’s TLs among families with married/resident fathers, but paternal incarceration was unrelated to maternal depression and youth’s TLs among families with non-married/resident fathers.

We also tested the temporal ordering among our constructs. Specifically, we examined whether the timing of risk (i.e., paternal incarceration, maternal depression) mattered for children’s telomeres, and we sought to rule out alternative hypotheses (e.g., children with shorter telomeres were by chance born into households with incarcerated fathers and depressed mothers). Age-9 paternal incarceration predicted youth’s age-9 and age-15 shortened TLs, and the timing of maternal depression was inconsequential to youth’s telomeres at both ages. Youth’s TLs did not predict paternal incarceration nor maternal depression in subsequent waves. Overall, these results supported the direction of effects in our hypotheses.

Discussion

Fifty-four percent of incarcerated people are parents with minor children.² Considering incarcerated people are more likely to be ethnic-racial minorities and men,² millions of ethnic-racial minority children are coping with paternal incarceration. Although associations between paternal incarceration and deleterious youth outcomes have been documented,²⁰ the process through which incarceration shapes youth’s well-being is under-studied. After controlling for individual, neighborhood, and school-level factors, our longitudinal analyses indicated that children who experienced paternal incarceration between ages 9 and 15 demonstrated greater accelerated TL shortening. Vitally, maternal depression mediated this link; mothers who experienced a partner’s incarceration also reported a greater risk for depression, and maternal depression predicted TL shortening among children. These findings illuminate how paternal incarceration impacts children’s development via biological, social, and ecological mechanisms.

In line with the EBD framework, maternal depression helped explain how paternal incarceration affected children’s telomeres. Considering the social and economic challenges facing them as single parents, mothers may have been upset over their partners’ incarceration or factors related to it. Maternal depression has been associated with stressful home environments for children^{30, 31} due to the social and economic challenges following paternal incarceration.^{41, 42} Notably, our results were more pronounced for households in which the father was married to or co-habited with the mother, supporting prior research.^{12, 42}

We were surprised that telomeres lengthened over time for most youth, and this increase in TL was larger than its standard error of measurement. Although telomeres are said to shorten with age, increases in TL have been found among children with short baseline TLs.⁴³ Whereas some scholars may attribute increased TL to measurement error, others have suggested that such increases are likely for those who transition from maladaptive to healthy contexts.⁴⁴ Most of the children in this sample were born into a socially disadvantaged group of unwedded and low-income parents; therefore, the children may have been born with shortened TLs. In addition, macro-economic events—the 2008 U.S. recession and the economic recovery that coincided with the age-9 and age-15 assessments—may have partly contributed to youth’s telomere lengthening as their economic conditions improved. Because such hypotheses are speculative, more research is needed to understand when telomeres lengthen over time.

In this sample, one-in-four Black and one-in-five multi-ethnic-racial children experienced paternal incarceration, whereas one-in-ten White youth did. Ethnic-racial inequalities and concentrated socially disadvantaged circumstances in ethnic-racial minority communities likely shaped the over-representation of ethnic-racial minority men in the criminal justice system.^{1, 18} In addition, most Black and multi-ethnic-racial participants lived in low-income and predominantly Black neighborhoods, which have been found to experience more punitive municipal policies relative to economically affluent and White neighborhoods.² However, our proposition regarding ethnic-racial bias contributing to these incarceration rates is speculative.

We did not find evidence that the harmful effects of paternal incarceration were specific to a particular ethnic-racial group, supporting extant findings.^{11, 12} However, our ethnic-racial groups varied significantly in socioeconomic status, marital status, and neighborhood settings, with more White households living in more favorable socioeconomic conditions than their Black and multi-ethnic-racial counterparts. These confounds potentially made it difficult for us to make reliable ethnic-racial group comparisons and, ultimately, likely masked some ethnic-racial group differences. It is also likely that youth across ethnically-racially diverse backgrounds respond more similarly to father loss to the criminal justice system. Nonetheless, more research is needed to distinguish ethnicity-race from other social processes (e.g., socioeconomic status) to specify its role in families’ criminal justice system involvement.

The present study had several limitations. First, we did not specify the rationale of incarceration, as prior research showed that children’s outcomes were unrelated to fathers’ incarceration for violent infractions.⁴⁵ Second, we assessed low-income mothers’ self-reported depression. Our findings may be stronger relative to a more general sample of mothers, as rates of maternal depression may be higher in our research participants than in a more general sample.⁴⁶ Third, other unmeasured variables associated with low-income, unwedded households (e.g., poor housing, financial stress, and unmet healthcare needs)⁴¹ may have also contributed to TL shortening. Fourth, due to the lack of nesting at the school-level, we could not account for school-level paternal incarceration, which has previously shaped children’s development.⁴⁷ Fifth, the association between TL at each time point was lower than expected for a relatively short period of time and may have been due to the assay or bias by environmental factors that occurred between waves. Finally, our effect sizes represented change scores across a six-year longitudinal span, making them small and difficult to interpret. Scholars should explore these hypotheses and consider these limitations in future research.

The present study also had several strengths. First, we used longitudinal data to understand how risk factors predicted changes in children’s cellular age. Second, although heritability is a strong predictor of TL,⁴⁸ our outcome of interest was the change in youth’s telomeres relative to their own average. This approach helped minimize effects from youth who started the study with shorter telomeres. Third, our results were robust after we accounted for youth’s self-rated health and internalizing symptoms; therefore, our findings cannot be solely attributed to the intergenerational transmission of depression. Fourth, we accounted for other forms of ACEs (i.e., maternal depression, parental separation, paternal substance/alcohol use, and multiple forms of father loss) to isolate the consequences associated with paternal incarceration. Among our ACEs, maternal depression was the most predictive of youth’s telomeres, which supports extant claims that proximal stressors are most linked to youth’s telomeres,^{49, 50} which is unsurprising as most children in our sample lived with mothers. Lastly, maternal depression did not predict subsequent paternal incarceration in our sensitivity analysis, which provides evidence that the two constructs are distinguishable types of ACEs. Future research should distinguish between ACEs to understand how ecological systems interplay and shape children’s development.

Ethnic-racial disparities continue to persist in chronic diseases. For instance, Black Americans are 70% and 30% more likely to struggle with diabetes and cancer, respectively, than White Americans. The fact that shortened telomeres are associated with these diseases^{9, 26} should urge scholars to consider children’s telomeres as possible mechanisms through which disparities in mass incarceration practices shape racial disparities in diseases for the following generation of Americans. In addition, we found that paternal incarceration predicted children’s TL shortening through maternal depression, which positions maternal mental health as a potential target for intervention. Toxic stress can be reduced when supportive adults facilitate children’s adaptive coping behaviors, thereby reducing their physiological responses to stress.⁶ In light of this information, practitioners should develop strategies that reduce risks to parental separation through combatting incarceration-based stigma (e.g., via counseling/support groups), promoting family visits (e.g., affordable transportation, community-based residential facilities, and developmentally appropriate family visiting spaces), and reducing fathers’ likelihood of re-entry (e.g., facilitate rehabilitation through education, job training, and employment following release).

In conclusion, an intergenerational association emerged between paternal incarceration and children’s health. Specifically, children with incarcerated fathers had shorter telomeres, which is a concern as shortened telomeres, in adult-based studies, have been associated with elevated health risks and in some cases predictive of negative health outcomes. Ethnic-racial disparities within rates of paternal incarceration illustrate how institutional discrimination can shape health inequalities among families. We caution that unnecessary father loss, including paternal incarceration as pretrial detainees or as falsely convicted individuals, comes at a cost to the well-being of children. Through family-focused strategies, we have the possibility to drastically reduce the consequences of incarceration that would otherwise contribute to ethnic-racial health disparities for the next generation; when it comes to children, if we can, then we should.