Parental Incarceration and Child Overweight: Results From a Sample of Disadvantaged Children in the United States

Abstract

Objectives:

Rates of childhood obesity and parental incarceration have been increasing in the United States since the 1970s. We examined whether parental incarceration was associated with child overweight at age 9 and whether that association differed by which parent was incarcerated.

Methods:

We ran cross-sectional logistic regression models predicting the likelihood of a child being overweight, conditional on whether the child’s mother, father, or both parents had ever been incarcerated. Our sample included non-Hispanic black (n = 1638) and non-Hispanic white (n = 504) children who were surveyed at age 9 in the Fragile Families and Child Wellbeing Study, a stratified random sample of new and mostly unmarried parents in US cities with populations >200 000. Children were born between 1998 and 2000.

Results:

Controlling for race, sex, parental socioeconomic status, maternal obesity, parental impulsivity, and child temperament, the odds of being overweight were significantly lower if a child’s mother was the only parent ever incarcerated (adjusted odds ratio [aOR] = 0.43; 95% confidence interval [CI], 0.22-0.83) or if both parents were ever incarcerated (aOR = 0.58; 95% CI, 0.38-0.87) compared with children whose parents were never incarcerated. Having a father who had ever been incarcerated had a comparatively meager and nonsignificant effect on the odds of a child being overweight (aOR = 0.81; 95% CI, 0.64-1.01).

Conclusions:

Our findings contribute to an emerging body of research suggesting that the consequences of parental incarceration for young children’s physical health may differ by whether a child’s mother or father has ever been incarcerated. Caution is warranted in generalizing the negative behavioral and mental health effects of parental incarceration to child physical health conditions, particularly for young children whose mothers have a history of incarceration.

The rate of incarceration in the United States has increased dramatically in the past 4 decades, from <200 per 100 000 persons in prison or jail in 1974 to >700 per 100 000 persons in prison or jail in 2012.¹ As a result, the number of young children with an incarcerated parent has also increased.² As of 2012, nearly 7% of children in the United States, amounting to approximately 5 million children, had lived with a parent who was incarcerated at some point since that child’s birth³; among black children in the United States, parental incarceration had risen to nearly 12% by the same year.³ A growing body of research has documented lasting consequences for these children on a range of psychosocial and behavioral outcomes, as well as on mortality.^4-6 Children of incarcerated parents perform less well in school, are at higher risk of learning disabilities and developmental delays, and are more likely to be placed in special education classes than children whose parents have never been incarcerated.^6,7 Research also shows that children of incarcerated parents are less emotionally well-adjusted and are at greater risk for antisocial behavior and delinquency than are children whose parents have never been incarcerated.^4,8

Less is known about how and whether the young children of incarcerated parents differ from children whose parents have never been incarcerated on physical indicators of health, such as overweight and obesity. Because the past 4 decades have also marked a dramatic increase in the rate of obesity among adults and children^9,10—from 1974 to 2008, adult obesity increased from 14%¹¹ to 34%¹⁰ and child obesity increased from 5% to 17%¹²—the question of whether parental incarceration predicts child body mass index (BMI) stands at the intersection of 2 pressing public health issues. In addition to being an important indicator of child health,¹³ obesity is known to persist strongly from childhood through adulthood,¹⁴ marking one pathway through which the incarceration of a parent may have lifelong physical health consequences for children.

As in much of the research on parental incarceration and child outcomes,^6,7 research on obesity among adolescents who have an incarcerated parent has hypothesized that effects are driven by the trauma of parental absence, finding that internalized stress may lead to an increased risk of overweight among adolescent girls.¹⁵ The mechanisms through which having an incarcerated parent affect child BMI, however, are likely to shift as children age and gain control over decisions about food intake and physical activity. For young children in particular, whose food consumption and physical activity are largely determined by caregivers, BMI may also be heavily influenced by shifts in caregiving resulting from parental incarceration. Therefore it is unclear whether the relationships among parental incarceration, stress, and BMI among teens¹⁵ or young adults¹⁶ can be directly generalized to young children.

Research further suggests that understanding the consequences of parental incarceration may require a focus not only on whether a parent was incarcerated, but also on which parent was incarcerated.^17,18 Although most studies of child outcomes after parental incarceration have focused on children with an incarcerated father,⁷ findings from the comparatively few studies considering child outcomes of maternal incarceration have been less consistent than findings on the child outcomes of paternal incarceration. Behavioral problems among children of incarcerated mothers, for example, appear to largely reflect disadvantage that precedes incarceration; maternal incarceration has even been found to decrease problem behaviors among non-Hispanic white children.¹⁷ Similarly ambiguous associations have been found between maternal incarceration and child mortality.⁵

Maternal incarceration also typically reflects a greater shift in caregivers than does paternal incarceration, because most children with an incarcerated father live with their mother, whereas only about one-third of children with an incarcerated mother live with their father.¹⁹ Children with incarcerated mothers, in contrast, are most commonly cared for by grandparents or other relatives.^19-23 If food intake and physical activity are more readily regulated by kinship caregivers than is the emotional experience of parental absence, one might observe positive outcomes of maternal incarceration on physical indicators of health among young children, such as overweight and obesity, without contradicting findings of negative psychological or behavioral outcomes. Although one study found that the association between parental incarceration and obesity in young children was fully attenuated by socioeconomic status and family stability, maternal incarceration and paternal incarceration were not differentiated in their analysis.⁶ The objectives of our study were to examine (1) whether parental incarceration was associated with the likelihood of a young child being a healthy weight, and (2) if any effect differed by whether the child’s mother or father was incarcerated.

Methods

We used a sample drawn from the Fragile Families and Child Wellbeing (FFCW) Study, a longitudinal survey of the new and mostly unmarried parents of 5000 children born between 1998 and 2000 in US cities with populations >200 000.²⁴ New parents were first interviewed shortly after childbirth, and mothers and fathers were then interviewed separately when the child was 1, 3, 5, and 9 years of age. For children not living with either parent, the primary caregiver was also interviewed. The survey covers a wide array of topics, including sociodemographic characteristics, family composition, and parent and child health.

Our analytical sample included all non-Hispanic black and non-Hispanic white children interviewed at age 9 in the FFCW Study for whom measurements of weight and height were available (n = 2142). We excluded Hispanic respondents and respondents of “other” race from our study because of the inability to distinguish these racial/ethnic groups from the immigrant experience more generally. Nearly half of the population of FFCW Hispanic children and more than half of children of “other” ethnicity who were interviewed at age 9 and for whom measurements of weight and height were available also had at least one foreign-born parent (compared with 5% of non-Hispanic white and non-Hispanic black children). Processes that influence incarceration and obesity are expected to differ among immigrant and non-immigrant families, given the legal challenges of immigrant status and the complex relationship between obesity and immigrant acculturation.²⁵

Few data were missing for any item. To account for missing data resulting from nonresponse, we coded missing values to zero and included an indicator to denote missing data. Although rates of nonresponse were high among birth fathers (more than 45% of birth fathers did not respond in year 9), approximately 75% of enrolled children and 70% of enrolled birth mothers remained in the sample by year 9. We decided not to impute missing data because of the large percentage of missing fathers and because missing data largely denoted parental nonresponse; therefore, values were plausibly not missing at random.²⁶ For example, education at baseline among fathers who did not respond during year 9 was significantly lower than education at baseline among fathers who responded during year 9 (P < .001), lending reason for concern about unobserved values that also differed among fathers who did and did not report during year 9. However, results were robust to supplemental analyses using list-wise deletion and multiple imputation; these results are available from the authors upon request. This study reports results from analysis of de-identified publicly released survey data and was considered exempt from institutional review board review.

Key Variables

Our independent variables of interest were indicators for a child’s mother (but not father) having been incarcerated, a child’s father (but not mother) having been incarcerated, and both of a child’s parents having been incarcerated. Mothers and fathers in the FFCW sample were asked at each interview about their own and their partner’s incarceration history; at year 9, nonparental primary caregivers were also asked about parental incarceration (full list of questions available from authors upon request). We calculated 2 measures of parental incarceration: one measure that captured the occurrence of any incarceration,⁷ and to ensure correct time-ordering of covariates, another measure that excluded incarceration before a child’s birth.¹⁷ Both measures produced substantively similar results. Because reports of incarceration in the FFCW data primarily come from mothers, information on whether incarceration occurred is likely to be accurate, but timing of incarceration is less clear. We therefore presented our results using the more inclusive measure of whether a parent was ever reportedly incarcerated.

The Centers for Disease Control and Prevention defines overweight among children aged 2-19 years as having a BMI ≥85th percentile of persons of the same age and sex, determined by using representative data of the US population of persons aged 2-19 years collected in various surveys from 1963-1965 to 1988-1994.²⁷ The BMI associated with each of these percentiles varies by sex and age (in months) until age 20.²⁸ Interviewers in each year of the FFCW Study directly measured weight in pounds and height in centimeters of enrolled children as part of the in-home assessment. These measurements were taken for fewer than half of the sample of children during years 3 and 5, but they were taken for 70% of the sample of children during year 9. Because obesity is known to persist from childhood through adulthood, the likelihood of a child who was at an unhealthy weight at an early age (eg, age 3) being classified as normal weight by age 9 is low.¹⁴

Covariates and Analysis

Our models included control variables expected to influence both parental incarceration and child obesity, including a child’s race and sex, as well as the education, income, and age of both the child’s mother and father at the time of the child’s birth. Education was assessed on a 4-point scale (1 = <high school, 2 = high school, 3 = some college or technical training, 4 = college or graduate degree). Income was assessed in dollars and transformed by using the inverse hyperbolic sine function (similar to a logarithmic transformation, but linear through the origin²⁹). A measure of parental impulsivity based on the dysfunctional impulsivity questions from Dickman’s impulsivity scale,³⁰ which is expected to affect both parenting behaviors and the likelihood of incarceration, was also included.^31,32 This measure was assessed via self-rating and partner-rating for both parents. Mothers reported on the child’s behavior at each interview by using items taken from the EAS Temperament Survey for Children³³; we included a summary measure of child temperament at age 1 to address concerns with the time ordering of child behavior and weight gain. This measure ranged from 0 to 1, with higher values indicating more positive child temperament. We included a continuous measure for the number of children in the mother’s and father’s household at year 9, because a larger family size is associated with lower levels of receipt of household resources among children.³⁴ To account for familial tendencies toward overweight, we included an indicator for whether a child’s biological mother was obese during any of the interviews in which maternal BMI was recorded (years 3, 5, and 9). Finally, we included in all models a fixed effect on child birth city, although in supplemental models not accounting for city we observed no substantive differences from our main findings.

We estimated a series of logistic regression models in which the outcome was whether a child was above a healthy weight, regressed on our 3 measures of parental incarceration and the controls described. In model 1, we controlled for race and sex; in model 2, we controlled for race, sex, maternal and paternal socioeconomic status, and maternal obesity; and in model 3, we controlled for race, sex, maternal and paternal socioeconomic status, maternal obesity, parental impulsivity, and child temperament. We report the results of Wald tests to affirm differences in the odds of our outcome between maternal and paternal incarceration.

Sensitivity Analyses

To affirm that our findings were robust to a more stringent test of time ordering, we ran a third specification of our models regressing overweight at age 9 on only new parental incarceration between the time the child was age 3 and age 9, controlling for previous parental incarceration, child overweight at age 3, and all controls presented in the main models. We conducted further sensitivity analyses by examining whether the results observed persisted when the outcome was defined as obesity (BMI ≥95th percentile) rather than overweight (BMI 85th to <95th percentile percentile), and whether we saw a converse relationship when the outcome was defined as normal weight (BMI 5th to <85th percentile), to ensure that any reduction in the odds of overweight was not due to an increase in underweight among children of incarcerated parents. We also examined whether results differed by sex. Supplemental models included control variables for whether a child received free breakfast or lunch at school, whether a child participated in athletic activities, indicators for shifts in the child’s primary caregiver at each survey year, and indicators of food insecurity, such as the receipt of Supplemental Nutrition Assistance Program (SNAP) benefits. These covariates were not significant predictors of child overweight and were excluded from the final models.

Results

Descriptive Statistics

Of the 2142 children in the study, the rate of maternal incarceration (n = 209, 9.2%) was substantially lower than the rate of paternal incarceration (n = 1135, 52.9%). The rate of overweight among children aged 9 in our sample was 39.6% (n = 848), and about half (52.9%, n = 1133) of mothers were obese (Table 1). The rate of overweight was lower among children whose mothers were ever incarcerated than among children whose mothers were never incarcerated. More than 40% of children whose parents had never been incarcerated were overweight; 39.9% of children whose father but not mother had been incarcerated were overweight; 33.5% of children for whom both parents had been incarcerated were overweight; and 27.5% of children whose mother but not father had been incarcerated were overweight. Rates of overweight among children whose mothers had ever been incarcerated were significantly lower than among children whose parents had no history of incarceration (P = .03).

Table 1.

Means and frequencies for selected variables, by incarceration status and by which parent was incarcerated, among 2142 children aged 9 years in the Fragile Families and Child Wellbeing Study, United States, 2007-2010^a

Variable	Full Sample (N = 2142)	By Which Parent Was Ever Incarcerated
		Neither Parent (n = 956)	Mother Only (n = 51)	Father Only (n = 977)	Both Parents (n = 158)
Weight status of child at age 9, no. (%)
BMI ≥85th percentileb	848 (39.6)	391 (40.9)	14 (27.5)	390 (39.9)	53 (33.5)
BMI <85th percentile	1294 (60.4)	565 (59.1)	37 (72.5)	587 (60.1)	105 (66.5)
Race/ethnicity of child,c no. (%)
Non-Hispanic black	1638 (76.5)	616 (64.4)	42 (82.4)	853 (87.3)	127 (80.4)
Non-Hispanic white	504 (23.5)	340 (35.6)	9 (17.7)	124 (12.7)	31 (19.6)
Sex of child, no. (%)
Female	1020 (47.6)	455 (47.6)	30 (58.8)	463 (47.4)	71 (44.9)
Male	1122 (52.4)	501 (52.4)	21 (41.2)	514 (52.6)	87 (55.1)
Weight status of mother at year 9, no. (%)
BMI ≥95th percentileb	1133 (52.9)	490 (51.3)	25 (49.0)	541 (55.4)	77 (48.7)
BMI <95th percentile	882 (43.8)	421 (46.2)	21 (45.7)	379 (41.2)	61 (44.2)
SES of mother at year 0
Age, mean (SD), y	25.1 (6.0)	27.4 (6.3)	26.9 (6.9)	23.0 (4.9)	23.8 (5.6)
Educationd	2.2 (1.0)	2.6 (1.0)	2.0 (1.0)	1.9 (0.8)	1.7 (0.8)
Income (arcsinh)d	10.6 (1.4)	11.0 (1.4)	10.5 (0.9)	10.2 (1.4)	10.1 (1.2)
No. of children in household, mean (SD)	1.3 (1.3)	1.2 (1.3)	1.6 (1.5)	1.4 (1.4)	1.4 (1.4)
SES of father at year 0
Age, mean (SD), y	27.9 (6.7)	29.8 (7.0)	29.4 (8.2)	26.0 (5.6)	27.1 (6.9)
Educationd	2.2 (0.9)	2.6 (1.0)	2.1 (0.9)	1.9 (0.7)	1.8 (0.7)
Income (arcsinh)d	10.8 (1.5)	11.1 (1.4)	10.6 (1.8)	10.5 (1.5)	10.6 (0.8)
No. of children in household, mean (SD)	1.0 (1.1)	0.9 (1.0)	1.1 (1.3)	1.0 (1.1)	1.0 (1.2)
Parental impulsivity scalee
Self-reported, mother	0.1 (1.1)	0.2 (1.0)	–0.2 (1.1)	0 (1.1)	–0.4 (1.2)
Partner-reported, mother	0 (1.1)	0.3 (1.0)	–0.5 (1.3)	–0.1 (1.0)	–0.7 (1.2)
Self-reported, father	0 (1.1)	0.3 (0.9)	0.1 (0.9)	–0.1 (1.1)	–0.3 (1.2)
Partner-reported, father	–0.1 (1.1)	0.4 (0.9)	0.2 (0.8)	–0.4 (1.1)	–0.6 (1.1)
Child temperament scale, reported by mother when child was aged 1 yearf	0.57 (0.13)	0.58 (0.12)	0.58 (0.13)	0.56 (0.13)	0.55 (0.14)

Abbreviation: BMI, body mass index; SES, socioeconomic status.

^aData source: Reichman et al.²⁴

^bUsing a binary indicator of whether children were above the “normal weight” category (BMI ≥85th percentile); for mothers, differentiates only between whether or not mothers were obese (BMI ≥95th percentile).

^cSample includes all non-Hispanic black and non-Hispanic white children aged 9 years in the Fragile Families and Child Wellbeing Study for whom weight and height were recorded. Processes that influence incarceration and obesity are expected to differ among immigrant and non-immigrant families, so respondents who were Hispanic or “other” race were excluded on the basis that more than half of these children had foreign-born parents.

^dEducation was assessed as a 4-category scale of degree completion. Income was assessed in dollars and transformed using the inverse hyperbolic sine function (similar to a logarithmic transformation, but linear through the origin).³¹

^eMeasures of parental impulsivity were based on the dysfunctional impulsivity questions from Dickman’s impulsivity scale.³² Values were transformed to have a mean of 0 and a standard deviation of 1. Higher values indicated more impulsivity.

^fMothers reported on the child’s behavior at each interview using items taken from the EAS Temperament Survey for Children.³⁵ This measure ranges from 0 to 1, with higher values indicating more positive child temperament.

Multivariate Results

The odds of a child being above a healthy weight were nonsignificantly reduced by having only a father incarcerated (model 3: adjusted OR [aOR] = 0.81; 95% CI, 0.64-1.01), but the odds of being overweight among children of incarcerated mothers were significantly lower than among children with similar characteristics whose parents had not been incarcerated. This reduction in the odds of overweight was significant for children whose mother was the only parent incarcerated (model 3: aOR = 0.43; 95% CI, 0.22-0.83) (Table 2) and for children whose mothers and fathers had both been incarcerated (model 3: aOR = 0.58; 95% CI, 0.38-0.87). Controls for parental socioeconomic status, parental impulsivity, child temperament, and maternal obesity did not significantly attenuate the effects observed. The difference in the odds of obesity among children whose mother but not father had been incarcerated compared with children whose parents had both been incarcerated was not significant (P = .43), whereas the difference in the odds of obesity among children who experienced any maternal incarceration compared with children who experienced only paternal incarceration was significant (P = .02).

Table 2.

Overweight among 2142 children aged 9 years in the Fragile Families and Child Wellbeing Study, by selected characteristics of children and parents and by parental incarceration, United States, 2007-2010^a

Variable	Model 1,b aOR (95% CI) [P Value]c	Model 2,b aOR (95% CI) [P Value]c	Model 3,b aOR (95% CI) [P Value]c
By which parent was ever incarcerated
Mother only	0.47 (0.25-0.90) [.02]	0.46 (0.24-0.89) [.02]	0.43 (0.22-0.83) [.01]
Both parents	0.68 (0.47-0.98) [.04]	0.67 (0.46-1.00) [.05]	0.58 (0.38-0.87) [.01]
Father only	0.85 (0.70-1.03) [.10]	0.84 (0.68-1.04) [.12]	0.81 (0.64-1.01) [.06]
Neither parent	1 (Reference)	1 (Reference)	1 (Reference)
Race/ethnicity of child
Non-Hispanic blackd	1.74 (1.36-2.22) [<.001]	1.46 (1.12-1.92) [.01]	1.43 (1.09-1.88) [.01]
Non-Hispanic white	1 (Reference)	1 (Reference)	1 (Reference)
Sex of child
Female	1.40 (1.17-1.66) [<.001]	1.39 (1.16-1.66) [<.001]	1.38 (1.15-1.66) [<.001]
Male	1 (Reference)	1 (Reference)	1 (Reference)
Obesity status of mother at year 9
Obese		2.46 (2.02-2.99) [<.001]	2.46 (2.02-3.00) [<.001]
Not obese		1 (Reference)	1 (Reference)
SES of mother at baseline
Age		1.02 (1.00-1.05) [.05]	1.02 (1.00-1.05) [.04]
Educatione		1.00 (0.88-1.13) [.99]	1.01 (0.89-1.15) [.83]
Income (arcsinh)e		1.02 (0.94-1.10) [.65]	1.03 (0.95-1.11) [.53]
No. of children in household		0.88 (0.81-0.96) [<.001]	0.88 (0.80-0.95) [<.001]
SES of father at baseline
Age		1.00 (0.98-1.02) [.82]	1.00 (0.98-1.02) [.79]
Educatione		0.85 (0.75-0.97) [.01]	0.87 (0.76-0.99) [.03]
Income (arcsinh)e		1.01 (0.94-1.08) [.86]	1.00 (0.93-1.08) [.97]
No. of children in household		1.10 (1.00-1.22) [.04]	1.10 (1.00-1.21) [.05]
Parental impulsivity scalef
Self-reported, mother			0.92 (0.84-1.01) [.07]
Partner-reported, mother			0.97 (0.88-1.06) [.45]
Self-reported, father			0.98 (0.90-1.08) [.72]
Partner-reported, father			0.95 (0.87-1.05) [.32]
Child temperament scale, reported by mother when the child was aged 1 yearg			0.55 (0.26-1.15) [.11]
Constant	0.50 (0.31-0.80) [<.001]	0.24 (0.08-0.76) [.02]	0.32 (0.09-1.11) [.07]

Abbreviations: aOR, adjusted odds ratio; SES, socioeconomic status.

^aData source: Reichman et al.²⁴

^bModel 1 includes controls for race and sex. Model 2 includes controls for race, sex, parental SES, and maternal obesity. Model 3 includes controls for race, sex, parental SES, maternal obesity, parental impulsivity, and child temperament.

^cP values were calculated based on Wald tests, with significance set at P < .05.

^dSample includes all non-Hispanic black and non-Hispanic white children aged 9 years in the Fragile Families and Child Wellbeing Study for whom weight and height were recorded.

^eEducation was assessed as a 4-category scale of degree completion. Income was assessed in dollars and transformed using the inverse hyperbolic sine function (similar to a logarithmic transformation, but linear through the origin).³¹

^fMeasures of parental impulsivity were based on the dysfunctional impulsivity questions from Dickman’s impulsivity scale.³² Values were transformed to have a mean of 0 and a standard deviation of 1. Higher values indicated more impulsivity.

^gMothers reported on the child’s behavior at each interview using items taken from the EAS Temperament Survey for Children.³⁵ This measure ranges from 0 to 1, with higher values indicating more positive child temperament.

Sensitivity Analyses

The direction of the results held when the outcome was defined as obesity (BMI ≥95th percentile) rather than overweight (BMI 85th to <95th percentile), although the level of significance varied. Children of incarcerated mothers also had significantly higher odds of being within the recommended range of BMI (BMI 5th to <85th percentile) compared with children whose parents had never been incarcerated. Although the use of year 3 BMI data restricted the sample size substantially (and, quite plausibly, nonrandomly), the pattern of results mirrored that of the main analysis (for the full model, paternal incarceration only: OR = 0.89; 95% CI, 0.51-1.55; maternal incarceration only: OR = 0.34; 95% CI, 0.13-0.88; both parents incarcerated: OR = 0.35; 95% CI, 0.05-2.58). Nonsignificance of the coefficient on new incarceration of both parents between years 3 and 9 reflected the small number of children in this category (n = 6). Finally, no evidence indicated that our findings differed by child’s sex (Wald tests for gender differences: maternal incarceration, P = .66; paternal incarceration, P = .12; both parents incarcerated, P = .73).

Discussion

To bridge research on the increasing rate of childhood obesity with research on the increasing rate of parental incarceration, we examined whether having an incarcerated parent was associated with overweight and obesity among young children. Although previous studies have demonstrated negative behavioral and academic outcomes among children of incarcerated parents,^6,7 our results suggest caution in extrapolating the direction of these effects onto physical health outcomes, such as being above a healthy weight. We found that parental incarceration was associated with lower odds of a child being overweight or obese, but only when the child’s mother had been incarcerated. Paternal incarceration alone had no significant effect on child body weight. This finding aligns with a growing body of research suggesting that parental incarceration is not a blanket experience, but may vary depending on whether a child’s mother or father is incarcerated.^5,17

Although research has suggested the stress of parental absence as a primary mechanism linking parental incarceration with negative child outcomes,^6,7 shifts in caregiving environments associated with parental incarceration may be particularly relevant for BMI among young children, whose food consumption and physical activity are largely regulated by adults. Because most children with an incarcerated father live with their mother but few children with an incarcerated mother live with their father,^19,20 any effect of shifting caregivers on overweight among young children of incarcerated parents should be concentrated among children with incarcerated mothers, mirroring the results we observed. The consequences of ongoing involvement of kinship caregivers on child well-being after maternal incarceration merit further study.

Our analysis was strictly associational, and was intended to lay the groundwork for future research testing the causal link between parental incarceration and physical health outcomes among young children. Establishing causation is a challenge in research on parental incarceration and child outcomes in the United States because despite survey data being well-suited for generating associations between self-reported conditions, most survey data are not well-suited for establishing causal effects. Time-use surveys could shed light on how parental incarceration affects the amount of time children spend eating, participating in physical activity, and being in the company of a responsible caregiver, but survey data are less useful for studying child physical health outcomes other than common chronic conditions such as obesity. Serious childhood illnesses (eg, diabetes) are relatively rare; as such, a nationally representative survey will rarely provide enough data for reasonable study. Equally concerning is the high likelihood of bias in the reporting of child health problems, health care contact (eg, hospitalizations), or health maintenance behaviors (eg, vaccinations), particularly among highest-risk families in which the potential for Child Protective Services (CPS) involvement is of greatest concern. Linked administrative data with reliably measured indicators of incarceration, child health conditions, and CPS contact would enable researchers to better understand how parental incarceration interacts with a child’s physical health.

Overweight is typically viewed as a negative health outcome,⁹ but one possible alternative explanation for our findings would be if normal weight in low-income households reflected caloric or nutritional deficiencies, given that child overweight and obesity are generally more common in low-income households than in higher-income households.³⁵ However, we ran supplemental models controlling for indicators of food insecurity (eg, receipt of SNAP benefits) and for consumption of high-calorie foods (eg, fast food and soda), and we found no evidence that undernourishment explained our results. Furthermore, the relationship between parental incarceration and primary caregiver reports of overall child health in our sample was consistent with the results found for overweight: maternal incarceration but not paternal incarceration was associated with higher odds of having a child reported as being in good overall health. That said, we are cautious about labeling maternal incarceration as a protective factor for child health; behaviors related to physical health, such as food intake and physical activity, may simply be easier for kinship caregivers to regulate than the emotional experience of parental absence.^5,7

Limitations

Our study had several limitations. First, because rates of obesity³⁵ and incarceration² are higher in socioeconomically disadvantaged populations, the FFCW data have the advantage of focusing on persons most at risk of being obese or incarcerated; however, our findings may not be generalized to children of higher socioeconomic status parents who have been incarcerated. Second, although few children of incarcerated mothers are placed in foster care,²⁰ the risk of CPS intervention is nonetheless a pressing issue among mothers who have been incarcerated, particularly incarcerated black mothers.^36-38 The FFCW Study asked parents to report contact with CPS, and in supplemental models, we included an indicator for reported CPS involvement. Although we found no effect of CPS contact on child overweight, reporting bias (the likelihood that concerned parents may not admit to CPS contact) may have threatened the validity of our measure.

Reporting bias may also be of concern for self-reports of incarceration in the FFCW Study, but having questions on incarceration that are phrased in multiple ways across multiple survey instruments and with redundancy across members of the same family unit lends more confidence on those variables. Of larger concern is that, as is standard in US research on incarceration, we were unable to reliably control for either duration or number of episodes of parental incarceration. We therefore could not assess whether longer or more frequent parental incarceration had larger effects on children than did shorter or singular episodes of incarceration. Measures of whether a child was living with grandparents or other relatives at the time of the FFCW interviews did not explain the associations we observed, nor did indicators of changes in a child’s primary caregiver between survey interviews; however, such indicators would not capture data on temporary stays with a relative during an episode of parental incarceration that did not coincide with the FFCW interview, or on extensive nonresident involvement of relatives in a child’s daily life after parental incarceration.

Conclusion

Although men in the United States are more likely to be incarcerated than are women, the percentage of mothers who have been incarcerated is large among vulnerable populations such as the FFCW sample: 11% of non-Hispanic black children and 9% of children of all races had a mother with a history of incarceration. Maternal incarceration nonetheless remains understudied, and constitutes a critical gap in the literature on child well-being in the era of mass incarceration. We emphasize the need for additional research on the pathways through which the consequences of maternal incarceration differ from the consequences of paternal incarceration, and on how parental incarceration differently affects child behavioral, mental health, and physical health outcomes.