PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of hhspaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Adopt Q. Author manuscript; available in PMC 2018 January 1.
Published in final edited form as:
Published online 2016 February 11. doi:  10.1080/10926755.2016.1149534
PMCID: PMC4774051
NIHMSID: NIHMS761473

Factors Associated with Adoption and Adoption Intentions of Nonparental Caregivers

Matthew D. Bramlett, Ph.D.1 and Laura F. Radel, M.P.P.2

Abstract

Data from the 2011–2012 National Survey of Children’s Health and the 2013 National Survey of Children in Nonparental Care were used to fit a multinomial logistic model comparing three groups to those who never considered adoption: those who ever considered, but are not currently planning adoption; those planning adoption; and those who adopted. Adoption may be more likely when the caregiver is a nonkin foster parent, a foster care agency was involved, and/or financial assistance is available. Those with plans to adopt but who have not adopted may face adoption barriers such as extreme poverty, lower education and being unmarried.

Keywords: adoption, foster care, relative care, grandparent care, national survey data

INTRODUCTION

Thousands of children in America are waiting to be adopted from foster care. The latest Adoption and Foster Care Analysis and Reporting System (AFCARS) report shows that from 2009 to 2013, there were approximately 400,000 children in foster care at any given time, with about the same number (250,000) entering and exiting foster care each year. More than 100,000 children are officially “waiting to be adopted” in each year, but less than 60,000 actually are adopted annually (US DHHS, 2014a). The number of children adopted from foster care has consistently been between 50,000 and 57,000 since at least 2002 despite the number waiting to be adopted being consistently twice as high or more (US DHHS, 2014b).

Researchers such as Leathers et al. (2012) have observed that, “little research has focused on factors that predict adoption or influence foster parents’ decision to adopt” (p.892). One factor that has been examined often using administrative data on children in foster care is the type of foster caregiver: kin to the child or not. Carnochan, Moore and Austin (2013) discuss numerous studies that found that kinship care is associated with longer stays in foster care. Berrick and colleagues (1994) found that kin foster parents are less likely than nonkin foster parents to adopt, and that a frequent reason was that the kin foster parent felt that the child was “already family” and did not feel a need to adopt to make that relationship more formalized, even though many intended to care for the child throughout childhood anyway (Berrick, Barth & Needell, 1994: p.57). Nationally, of adopted children in 2007 who had been adopted from foster care, 23% of adoptions were by relatives (Malm, Vandivere & McKlindon, 2011).

Many other factors have been found to differentiate adoptive foster families from nonadoptive foster families. Those that have been found to predict adoption include: characteristics of the child, such as younger age (Carnochan et al., 2013; McDonald et al., 2007; Snowden et al., 2008; Connell et al., 2006); characteristics of the birth family, such as whether the parental rights were relinquished (Carnochan et al., 2013); characteristics of the adoptive family, such as white race (Carnochan et al., 2013) and being married (Carnochan et al., 2013, McDonald et al., 2007;); and system characteristics such as receipt of Temporary Assistance for Needy Families (TANF) or subsidies (McDonald et al., 2007).

Factors that have been found to be associated with lower likelihood of adoption include characteristics of the child such as black race (Carnochan et al., 2013; McDonald et al., 2007; Snowden et al., 2008); characteristics of the birth family, such as whether the child was physically or sexually abused (Carnochan et al., 2013; McDonald et al., 2007; Connell et al., 2006); and system characteristics such as negative caseworker attitudes (Carnochan et al., 2013) and higher number of prior placements (Carnochan et al., 2013; McDonald et al., 2007). Yampolskaya et al., (2014) suggest that older children with a history of being abused have very low odds of being adopted.

Physical and mental disability have shown inconsistent effects in prior research. McDonald and colleagues (2007) found that emotional problems and mental retardation were related to longer waits for adoption, while other diagnosed conditions showed a similar but weaker relationship. Similarly, Connell et al. (2006) found that diagnosed mental conditions were associated with lower likelihood for adoption while disability had no effect. Some research has shown that physical disability is associated with higher odds of adoption while emotional problems are associated with lower odds of adoption (Snowden et al., 2008). Leathers et al. (2012) showed that externalizing behavior problems such as oppositional or aggressive behavior were negatively associated with adoption while internalizing problems such as depression and anxiety had no effect. Zill and Bramlett (2014) showed that diagnosed ADHD, but not depression/anxiety, behavior/conduct disorder, or receipt of mental health care, was significantly more likely among children adopted from foster care than among children in foster care, but in that kind of comparison it is possible that the ADHD was not diagnosed until after the adoption or developed after the adoption.

While most research examining factors associated with adoption examine children available for adoption who did and did not get adopted, some researchers have approached the issue from the opposite direction by asking adoptive parents what motivated them to adopt. Vandivere, Malm and Radel (2009) report the most frequently cited reason adoptive parents gave for adopting was to provide a permanent home for a child in need, which was a motivating factor for the adoptive parents of 81% of all adopted children ages 0–17 in 2007, and for adoptive parents of 86% of children adopted from foster care. It can be argued that all children in foster care, or even all children living apart from any biological or adoptive parents, are in need of a permanent home. Other reasons given for adopting include: to expand the family (61%), because of infertility (39%), or because they wanted a sibling for a child (24%) (Malm, Vandivere & McKlindon, 2011). Reasons given for specifically choosing to adopt from foster care rather than from another country or through a private domestic adoption include lower cost (60%), faster process (28%), wanting a special needs child (24%), and wanting an older child (14%) (Malm, Vandivere & McKlindon, 2011).

We have chosen here to examine adoption among all children living without a biological or adoptive parent present in the household (i.e., all children in nonparental care). Most of these children are living with grandparents or foster parents, some with other relatives (e.g. aunts, uncles, older siblings), and a few are living with nonrelatives such as godparents, family friends, coaches, or others. We include all children in nonparental care in our analysis because the populations of children in relative care outside the child welfare system and inside the child welfare system are indistinct. That a child is in relative foster care may depend less on the child’s circumstances and need for a permanent family than on state policy choices as well as whether relatives or child welfare authorities were first to step in to address the parent’s unavailability.

The present authors have previously compared the health and well-being of types of children who have and have not been adopted: stepchildren who have and have not been adopted by their step-parent (Bramlett, 2010), children living with relatives who did and did not adopt them (Radel, Bramlett & Waters, 2010), and children in foster care and adopted from foster care (Zill & Bramlett, 2014). Each of these analyses was limited in not being able to distinguish, among caregivers who had not adopted, those who wanted to adopt or were in the process of adopting the children in their care – i.e., those most likely to adopt among the pool of potential adopters. The 2013 National Survey of Children in Nonparental Care (http://www.cdc.gov/nchs/slaits/nscnc.htm) includes the necessary data to examine the factors associated with both adoption and adoption intentions for a national sample of children in noninstitutionalized nonparental care (i.e., children in households with no parents present).

McDonald, et al. (2007) partitioned the adoption process into discrete elements of placement and finalization to better predict speed of adoption. We are partitioning the earlier adoption decision into elements of considering, planning, and carrying out the adoption to better predict characteristics associated with different stages of deciding to adopt and adopting. Two national studies have examined the frequency which with those who say they have considered adoption actually do so. Both found that relatively few follow through and take steps to adopt. Jones (2008) using data from the National Survey of Family Growth found that 36 percent of ever married women in 2002 had considered adopting, but only 17 percent had taken steps to adopt and only 1.7 percent had actually adopted a child. Harris Interactive and the Dave Thomas Foundation for Adoption (2007) similarly found in a national opinion survey that many more people reported having seriously considered adoption than had acted to do so.

The present study contributes to the “Finding adoptive families” aspect of this current special issue of Adoption Quarterly by examining characteristics of the child, the caregiver, and the living arrangement that are associated with adoption and the caregiver’s adoption intentions using a large-scale, population-based national survey of children in nonparental care.

METHODS

Data

Data were drawn from two national surveys conducted by the National Center for Health Statistics (NCHS): the 2011–2012 National Survey of Children’s Health (NSCH), a nationally-representative survey of households with children, and the 2013 National Survey of Children in Nonparental Care (NSCNC), which re-interviewed almost 1,300 households identified as nonparental care households in the NSCH, including foster care, grandparent care, and other households with no parents present. Both surveys were modules of NCHS’ State and Local Area Integrated Telephone Survey (SLAITS). NSCH was sponsored by the Health Resources and Services Administration’s Maternal and Child Health Bureau, while NSCNC was sponsored by the Department of Health and Human Services’ Office of the Assistant Secretary for Planning and Evaluation, with supplemental funding from the Annie E. Casey Foundation.

NSCH was a random-digit-dial landline and cell telephone survey that interviewed 95,677 households with children throughout the 50 states and the District of Columbia. The NSCH sample is nationally representative of children aged 0 to 17 years in households in the United States in 2011–2012. One child in each household with children was randomly selected to be the target of the NSCH interview. Respondents were adults in the household who were knowledgeable about the child’s health, usually the child’s mother.

NSCNC was a follow-back survey 1–2 years after the NSCH for children who had been living in households with no parents present and were ages 0–16 in 2011–2012 (i.e., during the administration of the NSCH). Follow-back interviews were conducted via telephone with a current caregiver of the child, in some cases the parent who had reunited with the child since the NSCH interview. The NSCH had a 51% cooperation rate among eligible households but a 23% overall response rate (partly due to the inclusion of cell-phone sample to maximize coverage of the population), and NSCNC had a 52% completion rate among eligible households 1–2 years later. Weighting adjustments were applied such that the population estimated by the weighted sample of completed NSCNC interviews matched that of the pool of eligible households demographically. This dramatically reduced estimated nonresponse bias such that remaining bias in weighted estimates was smaller than sampling error. More information about NSCH and NSCNC may be found at: http://www.cdc.gov/nchs/slaits.htm or by referring to the associated documentation published by the SLAITS program (CDC, 2013; CDC, 2014).

The analysis sample for the current study included those children in the NSCNC who were not living with a biological parent at NSCNC (i.e., had not been reunited with a parent between interviews). This restriction placed 1,160 of the 1,298 NSCNC children in the analysis sample.

Statistical Analysis

Weighted estimates were calculated using SUDAAN to account for the complex sample design of the NSCH (RTI, 2008). A multinomial logistic regression model was fitted that compares 3 levels of adoption intentions and behavior, relative to those who never considered adoption: 1) those who ever considered, but are not currently planning to adopt; 2) those currently planning to adopt; and 3) those who adopted the child in the 1–2 years between the NSCH and NSCNC interviews. Further analysis examined the caregiver-reported reasons that discouraged those without current plans to adopt from adopting.

Measures

The dependent variable is derived from multiple questions on the survey. Children whose caregivers responded that they and/or their spouse/partner had adopted the child, or who lived with adoptive parents at NSCNC interview, were coded to “adopted.” Caregivers were asked whether they had ever considered adopting the child and whether they were currently planning to adopt the child and these questions were used to categorize children into the remaining groups.

The multinomial model includes measures from various domains: characteristics of the child, of the household, and of the caregiver; aspects of the nonparental care living arrangement; and indicators of caregivers’ parenting characteristics. Additionally, a continuous variable measuring the number of days between the NSCH and NSCNC interviews was included as a control variable.

In addition to child demographics (age, sex and race/ethnicity), other child characteristics include: overall health (rated excellent or very good versus good/fair/poor); whether the child received mental health care in the 12 months prior to the NSCNC interview; whether the child has Attention-Deficit/Hyperactivity Disorder (ADHD); whether the child has special health care needs (any of five health care consequences resulting from a chronic health condition: (1) ongoing need for prescription medications; (2) ongoing need for specialized therapies; (3) ongoing need for more health care services than most children the same age; (4) treatment for a behavioral, developmental or emotional problem; and/or (5) activity limitation); and eight measures of adverse family experiences (AFEs) the child may have had: whether (1) the household had often found it hard to afford basics and whether the child had ever (2) experienced the death of a parent, (3) experienced the divorce or separation of a parent, (4) experienced the incarceration of a parent, (5) witnessed violence in the home, (6) experienced or witnessed violence in the neighborhood, (7) lived with a mentally ill person, or (8) lived with a substance abuser.

Household socioeconomic characteristics include income relative to Federal Poverty Level (FPL), in four categories (0–50% or “very poor,” 50–100% or “poor,” 100–200% or “near-poor,” and more than 200% or “not-poor”), and education of caregiver respondent and spouse (if present) – whichever is higher – in two categories (less than or equal to high school versus more than high school).

Caregiver characteristics include type of nonparental care (foster kin, foster non-kin, nonfoster grandparent care, other nonfoster nongrandparent); caregiver age; caregiver marital status (married versus never or formerly married); caregiver employment status (indicators that the caregiver or spouse is employed full-time and that the caregiver or spouse is retired); caregiver’s health insurance type (private/employment-based, public or uninsured); whether the caregiver owns the home or not; whether either the caregiver or spouse is not in excellent or very good overall health, whether either the caregiver or spouse is not in excellent or very good mental health, whether the caregiver is depressed, and whether the child is of a different race or ethnicity than the caregiver.

Aspects of the living arrangement include whether Child Protective Services (CPS) or a foster care agency was involved in placing the child in the living arrangement; whether the caregiver received any money for caring for the child; whether the caregiver felt s/he had been very well prepared to care for the child; and whether the caregiver felt that information about adoption and custody issues and legal assistance had been difficult to obtain. Parenting characteristics include whether the caregiver felt that s/he had a good understanding of the child and whether the caregiver usually or always felt at least one form of parenting stress (anger at the child, feeling bothered at the child’s actions, or feeling that the child is harder to care for than normal).

Child sex, race/ethnicity, overall health, ADHD, special health care needs, AFEs, and parenting stress are measures drawn from NSCH (and some values may have changed between surveys). The remaining covariates are drawn from (and measured at) NSCNC. Including all these measures in the model, many with small amounts of missing data, results in a large percentage (25%) of cases being dropped from the model due to nonoverlapping missing data on one or another variable. To counter this, covariates were dropped from the model in descending order of p-value (i.e., least significant effect dropped first), if three conditions were met: 1) the covariate had a nontrivial amount of missing data, 2) the covariate had never been significant in any model permutation examined, and 3) removing the variable did not substantially change other results in the model. The final model that resulted from this process had 12.8% of the sample with missing data for any model covariate. (Dropping ADHD, which is significant in the model but has 5% missing data, would reduce the missing data in the model to 7% of the sample but since dropping this variable had no appreciable effect on the magnitude or direction of the other effects in the model, ADHD was retained). Characteristics included in the final model are shown in Table 1, while characteristics dropped from the model because of nonsignificance and missing data are shown in Table 2.

Table 1
Means and Percentages of Model Covariates by Level of Adoption Intention and Behavior
Table 2
Percentages of Additional Preliminary Model Covariates* by Level of Adoption Intention and Behavior

RESULTS

Overall, of children living in nonparental care at the time of the NSCH interview, 8.8% were living with adoptive parents (i.e., had been adopted) by the time of the NSCNC interview 1 to 2 years later and an additional 23.6% had caregivers with current plans to adopt them. In 20.9% of cases, caregivers had considered adopting but had no current plans to adopt the child and for 36%, caregivers had never considered adoption (the remaining 10.7% were living with biological parents at the time of the NSCNC interview – the above percentages, calculated among those not reunited with biological parents, were 9.9% adopted, 26.4% with caregivers planning to adopt, 23.4% with caregivers who considered adoption but were not planning adoption, and 40.3% with caregivers who had never considered adoption).

Table 1 shows means and percentages for model covariates, for the full analytic sample and by categories of the dependent variable. It is possible to read the four categories of the dependent variable, from left to right in this table, as an ordinal scale with ascending probabilities of adoption (culminating in 100% probability in the “already adopted” column). One model domain stands out (caregiving placement characteristics) because the “already adopted” group significantly differs from all other groups on almost all the variables in that domain. One column also stands out in that those children whose caregivers plan to adopt significantly differ in many ways from the other groups, notably that the children are much more likely to have special health care needs and to live in very poor households and their caregivers are much less likely to own their homes or live in households that include any retired people and are more likely to feel that they have a good understanding of the child.

Table 2 shows percentages, for the full analytic sample and by categories of the dependent variable, for variables that were dropped from the model. Since all were dropped from the model from a combination of both a) missing data and b) high p-value (very low significance), it is not surprising that all of the comparisons in Table 2 are nonsignificant. However, nonsignificant findings can be important, and what Table 2 shows is that certain characteristics are –perhaps surprisingly – not associated with adoption intentions or adoption. Specifically, caregiver insurance type, caregiver and child mental health issues, transracial/transethnic placement, and many adverse family events that may have contributed to the child’s placement in nonparental care do not seem to impact whether a nonparental caregiver considers, plans or carries out the adoption of the child in their care.

Table 3 shows the multinomial logistic regression results comparing each of three subgroups of children to children whose nonparental caregiver never considered adoption: those whose caregivers have considered adoption but are not currently planning to adopt, those whose caregivers are currently planning to adopt, and those whose caregivers adopted the child between the NSCH and NSCNC interviews. Higher odds ratios show the higher likelihood of the outcome relative to the reference group. Ever considered, not currently planning to adopt—Factors associated with higher odds of considering but not planning adoption include ADHD, nonkin foster care status and having experienced difficulty obtaining information about adoption or custody issues. Factors associated with lower odds of considering but not planning adoption include currently married status and having experienced difficulty obtaining legal assistance. The child’s non-Hispanic black race/ethnicity, kin foster care status and either the caregiver or caregiver’s spouse having less than very good overall health were marginally significant (i.e., significant at the p<0.10 level but not at the 0.05 level) with a lower likelihood of a caregiver having considered adopting .

Table 3
Multinomial Logistic Regression Model of Adoption Intentions & Behavior

Currently planning to adopt (relative to those who never considered adoption)

Factors associated with higher odds of planning adoption include the child being of Hispanic race/ethnicity (relative to non-Hispanic white), having special health care needs, or having lived with a substance abuser; lower household education; nonkin foster care status; having experienced difficulty obtaining information about adoption; and having a good understanding of the child. Factors associated with lower odds of planning adoption include older child age, poor or near-poor household income status (relative to not-poor), currently married status, home ownership, and either the caregiver or caregiver’s spouse having less than very good overall health. Ever having experienced parental separation or divorce and older caregiver age were marginally significant.

Already adopted (relative to those who never considered adoption)

Factors associated with higher odds of adoption include foster care agency or CPS involvement in the child’s placement in the home and receipt of payment for caring for the child. The child’s non-Hispanic other race/ethnicity, either the caregiver or caregiver’s spouse (if present) being retired, and having experienced difficulty obtaining information about adoption were marginally significant. Factors associated with lower odds of adoption include the child’s older age, excellent/very good overall health status, and ever having lived with the mentally ill; kin foster care status; and either the caregiver or caregiver’s spouse having less than very good overall health. Difficulty obtaining legal assistance and having a good understanding of the child were marginally significant.

Some of the strongest effects in table 3 include that children in kin foster care were much less likely to have already been adopted (OR of 0.03) and children in nonkin foster care were more likely to have caretakers who had either once considered adoption (OR 13.08) or had current plans (OR 11.36) to adopt. Additionally, children were much more likely to have been adopted already if the CPS agency had been involved in their placement (OR 5.75) or the caregivers received payment for caring for the child (OR 3.12). Having special health care needs was associated with a much higher likelihood of current adoption plans (OR 5.57) while factors associated with a particularly low likelihood of current adoption plans included low income (OR of 0.20 for poor caregivers and 0.10 for near poor caregivers), married caregivers (OR 0.24), caregivers who were homeowners (OR 0.19) and poor caregiver health (OR 0.30).

Some factors were significantly associated with more than one level of adoption intention or behavior. Older child age was associated with lower odds of both adoption and planning to adopt. Kin foster status was associated with lower odds of adoption or considering adoption while nonkin foster status was associated with higher odds of considering or planning adoption. Being married was associated with lower odds of considering or planning to adopt. Less-than-excellent or very good overall health on the part of the caregiver or caregiver’s spouse (if present) was associated with lower odds for all three outcomes; while having had difficulty with obtaining information about adoption or custody issues was associated with higher odds for all three outcomes.

Factors not associated with adoption intentions or behavior at any level, beyond those dropped from the model and shown in Table 2, include child sex, caregiver preparation, and parenting stress. The child having experienced parental separation or divorce, caregiver employment and caregiver age each had only one effect that was marginally significant.

Table 4 shows the reasons given for why those not currently planning to adopt were discouraged from adopting the child in their care. The most common reasons given for not planning to adopt were financial concerns and issues related to the birth parents (not relinquishing parental rights, strong parent-child attachment, parental discouragement) while the least common reasons given were child issues such as age, health, transracial status or behavior problems. Two thirds of nonparental caregivers either had another unspecified reason for not adopting (42%) or reported that there were no discouraging factors (27%).

Table 4
Factors that Discouraged Nonparental Caregivers from Adopting

DISCUSSION

Previous research on step-parent adoptions, relative care adoptions and adoptions from foster care (in each case comparing children adopted or not adopted) showed that adopting households tend to have comparable or more favorable socioeconomic attributes than nonadopting households. Foster and step adopters had higher household income and education compared to foster and step nonadopters, while relative adopters and nonadopters did not differ in education and income (Zill & Bramlett, 2014; Bramlett, 2010; Radel, Bramlett & Waters, 2010). As the current sample includes foster care and relative care households but not step-parent households, we expected to see a weak effect of higher education and income among adopters, relative to nonadopters (the foster effect diluted by the relative care non-effect). The results of the multinomial model, at first glance, would seem to show the expected effect for income and the opposite effect for education: adoption plans are less likely among the poor and near-poor than among the not-poor and are more likely among households with lower education than those with any college education, relative to those who never considered adopting.

However, income and education are only significant for predicting adoption plans and are not significant for predicting actual adoption, relative to never having considered adoption, specifically. The comparison group differs, potentially explaining the different income effect from that found in prior studies (that the already-adopted group does not have higher income than the group least likely to adopt). As Table 1 shows, household income and education did not significantly differ between children whose caregivers had never considered adoption and children who had been adopted. But both groups significantly differed from those whose caregivers had current plans to adopt, who had a significantly greater proportion in the “very-poor” income category (Table 1). And this group, with lower income, would have been included in the “had not adopted” group if the comparison had been done as in the prior studies, comparing the adopted to the not adopted. This conclusion highlights a limitation in analyses comparing types of children who have and have not been adopted, without distinguishing between those whose caregivers want to adopt and those whose caregivers have never considered adoption.

This finding also highlights the difference between adoption behavior and adoption intentions. Those children who have been adopted were less likely to live in very poor households; those whose caregivers plan to adopt them, but have not completed it yet, were almost three times as likely to live in very poor households, as seen in Table 1. This might indicate that some households intend to adopt but are unable to hurdle barriers to adoption that those with more resources are able to hurdle. This seems even more likely given that receipt of payment for caring for the child is associated with adoption (Table 3) and financial concerns was one of the most-endorsed reasons discouraging nonadopters from adopting (Table 4). The NSCNC sample includes many nonfoster caregivers who would not be eligible for subsidies that foster parents can receive, which may contribute to low adoption rates for those not in foster care.

The marital status effect may reflect a similar situation, in that the finding that unmarried caregivers are more likely to consider adoption or have current plans to adopt – but not necessarily to actually adopt – may be the result of barriers to adoption, in this case preferences to place children with married caregivers rather than unmarried caregivers. Although the traditional preference for married adoptive parents is not as strong as it once was and roughly one-third of children adopted from foster care are adopted by single adoptive parents (US DHHS, 2014a), some states continue to favor married couples over single persons as adoptive parents (Wilson & Wilcox, 2006; Associated Press, 2011). Private adoption agencies are generally not bound by state preferences for children in foster care and can and sometimes do restrict adoption by unmarried caregivers (Gardino et al., 2010).

There was a somewhat discrepant finding regarding child age, in that the model showed significantly lower likelihoods of adopting or planning to adopt for older children, but child age was also one of the least-endorsed reasons discouraging caregivers from adopting. A potentially lengthy adoption process makes little sense for older children if they are going to be adults or almost adults by the time the adoption is finalized. It is also likely that older children have more established social and emotional ties to their parents that they and their caregivers are not eager to dismiss by terminating parental rights, particularly if the caregiver is a relative. Kin caregivers may also think of this issue as “parental ties” rather than “child’s age” as being the factor discouraging the adoption. To investigate whether the age effect only applies to the oldest children, we estimated the model separately for teenagers (ages 13–17) and preteens (ages 1–12) (results not shown). In the teen model, the age effects were considerably stronger than in the full model (with ORs of 0.24 and 0.52 for currently planning to adopt and adoption instead of 0.81 and 0.85 respectively). In the preteen model, the age effects were weaker than in the full model (ORs of 0.83 and 0.98). Future research could consider modeling the effect of age as a curvilinear relationship to account for this issue.

Caregiver health was one of the few covariates that was significantly associated with all three outcomes, in that considering adoption, planning adoption and adopting are all less likely if either the caregiver or caregiver’s spouse (if present in the household) has less than excellent or very good overall health. The other covariates that were associated with all three outcomes were difficulty obtaining information about adoption and custody issues, and kin/nonkin foster status.

Difficulty obtaining adoption information and legal assistance were included in the model as attributes of the placement process that could have influenced adoption decisions and intentions. The model suggests that those who had difficulty obtaining information about adoption were more likely than those who found such information easily to consider or plan adoption (and to adopt as well, although that effect was not significant at the 0.05 level). This counter-intuitive finding is probably reflecting that those who do not wish to adopt are not actively looking for adoption information. Table 1 shows that among those who adopted, adoption information was much more likely to be easily obtained than difficult to obtain (75% versus 19%, respectively – a much larger difference than among those considering or planning adoption) but that the proportion of adopted children whose caregivers found that information hard to obtain was still three times as high as for those whose caregivers never considered adoption – explaining the model effect – because most of the nonadopters had presumably not tried very hard to obtain it. Thus, this variable should be considered a control variable rather than a predictor.

Kin/nonkin status of foster parents also had consistent effects, although not always significant at the 0.05 level: kin foster parents are less likely to adopt or consider adoption and nonkin foster parents are more likely to consider or plan to adopt. This finding is consistent with prior research (Carnochan et al., 2013; Berrick, Barth & Needell, 1994). Berrick, Barth and Needell (1994) indicated that kin foster parents are older, poorer, in poorer health, and more likely to be single than nonkin foster parents; our model controlled for all those attributes and still found that kin foster parents are less likely to adopt their foster children. Berrick, Barth and Needell (1994) found that the most frequently-cited reason among kin foster parents for not adopting was that they felt they were “already family” (p.57) and thus did not need to formalize that status with adoption. This, if true, may highlight a policy conundrum – there is a preference to place foster children with relatives and a preference to have foster parents adopt the child (Bussiere, 1998), and they may be contradictory goals to some extent. This is a key reason why the title IV-E kinship guardianship assistance program was created by the Fostering Connections to Success and Increasing Adoptions Act of 2008 (P.L. 110-351), to provide another financially supported option for permanency for relatives who planned to care for the child until adulthood but who were not interested in adoption (Stoltzfus, 2008). Future research is needed to determine whether guardianships are equal to adoptions in achieving stable, long-lasting permanency for children.

The relationship between a child’s mental health and his or her odds of being adopted is not clear. Analysis comparing adopted children to nonadopted children found that mental health issues such as learning disability, developmental delay, and emotional or behavioral problems are more prevalent among the population of adopted children (Bramlett, Radel & Blumberg, 2007). Specific to the population of children in relative care, Radel, Bramlett and Waters (2010) showed that developmental delay was more prevalent among children adopted by relatives than among unadopted children in relative care, while mental health care, learning disability and behavior/conduct problems did not differ between the groups. Specific to the foster care population, Zill and Bramlett (2014) showed that ADHD was more prevalent among children adopted from foster care than among children in foster care, while mental health care, depression/anxiety and behavior/conduct problems did not differ between the groups. Others have indicated that mental health diagnoses and emotional or behavioral problems are associated with lower odds of being adopted from foster care (Connell et al., 2006; Snowden, Leon & Sieracki, 2008; Leathers et al., 2012). Mental health conditions may operate differently from physical disabilities as some studies have found physical disabilities increase the probability of adoption (Snowden, Leon & Sieracki, 2008; Akin, 2011). This may factor into findings by Malm, Vandivere and McKlindon (2011) who demonstrate that among children adopted from foster care, 24% had foster parents who were motivated to adopt the child specifically because they wanted a special needs child (which could mean children with either physical or mental health conditions, or both).

If analyses of all adopted children (Bramlett, Radel & Blumberg, 2007) or of children adopted from foster care (Zill & Bramlett, 2014) show a higher percent of certain mental health issues than for nonadopted children, while mental health issues also make foster children less adoptable (Connell et al., 2006; Snowden, Leon & Sieracki, 2008; Leathers et al., 2012), it suggests that the background and process of being adopted (i.e., the conditions that precipitated the child being available for adoption and the uncertainty associated with changing caregivers, perhaps more than once) contribute to mental health issues (an eminently reasonable conclusion). It is also possible that some children with mental health conditions while in foster care do not get diagnosed with those conditions until after they are adopted. Research has suggested that adoptive parents may compensate for not being the child’s natural parents by seeking the best medical care possible (Kirk, 1984; Case & Paxson, 2001; Bramlett, Radel & Blumberg, 2007). Further, mental health issues may develop or intensify as children age, so there may be reason for diagnosis after adoption of problems that do not manifest until later. ADHD in particular tends to get diagnosed once children go to school and might not initially be diagnosed in younger children, so depending on age at adoption, the diagnosis might come after the adoption. A recent report from NCHS indicates that only one-third of children with ADHD were diagnosed before age 6 (Visser et al., 2015). We included measures of the child’s ADHD, receipt of mental health care, and whether they had ever lived with someone who was mentally ill as well as indicators of caregiver depression and less than very good caregiver/spouse mental health. We found that most of these did not predict adoption intentions or adoption in our model, while having ever lived with a mentally ill person was associated with lower odds of adoption, and having ADHD was associated with greater odds of considering adoption. Obviously, more research is needed to address this issue.

For those caregivers not planning to adopt, the reasons given why adoption was discouraged tended to focus on issues related to cost or the birth parents, notably the failure to terminate or relinquish parental rights and parental discouragement of the adoption (McDonald et al. (2007) showed that termination of parental rights can speed the adoption process substantially.) Issues related to the child, such as older age, transracial status, and behavior problems or juvenile justice system involvement, were the least-endorsed reasons. However, the majority of nonadopting caregivers did not endorse any of the reasons assessed in the survey, suggesting that there are many unique reasons why nonparental caregivers do not adopt the children in their care. Some respondents indicated that adoption was not necessary but did not specify why this was the case; others responded with some version of “the parents are still in the child’s life” or with an expectation or hope that the child would be reunited with parents, while a few indicated that the current situation was working and did not need to be “fixed.” This issue warrants exploration in future research.

We found that many of the issues that kin raise as discouraging them from adoption relate to the parents rather than the child. Thus, a key issue for ensuring permanence will be resolving mixed feelings of kin caregivers toward the birth parent(s) and making adoption an acceptable choice that does not feel like they or the child are giving up on or betraying the parent. Merritt (2008) showed that children’s feeling toward adoption by foster parents or kin developed over time, with more expressing an adoption preference as time went on. Treating adoption intentions as a process rather than a point in time choice may be helpful in increasing adoptions among this segment of the child welfare population.

Limitations

Findings from NSCH and NSCNC are based on parents’ experiences and perceptions. Information provided about health status and health care was not verified with health care professionals. Despite weighting adjustments to minimize nonresponse bias and evidence to suggest that remaining estimated biases tend to be smaller than sampling error (CDC, 2014), the low response rate means that bias resulting from nonresponse cannot be completely ruled out.

Largely due to the state-level sample design of the NSCH, which sampled each state and the District of Columbia in approximately equal sample sizes, design effects for national analyses tend to be large and confidence intervals around odds ratios are sometimes quite wide, even when odds ratios are statistically significant. Therefore, comparisons of the magnitudes of effects should include consideration of these confidence interval widths and may not be appropriate.

The most prevalent reasons – other than “other reason” – for not planning to adopt applied to fewer than 12% of children whose caregivers were not planning to adopt, while almost 42% had an “other reason.” Respondents who answered “other reason” were asked to specify, and most provided a reason that did not match existing categories. The verbatim responses included few related to issues with the child being the factor that discouraged adoption. This supports the conclusion based on Table 4 that issues with the birth parents were more prevalent than issues with the child as reasons for not adopting. The large number of unique responses provides a challenge in analyzing reasons for not adopting.

Despite these limitations, the authors know of no other data source than NSCNC that includes a large-scale, population-based national sample of all noninstitutionalized children living in nonparental care – i.e., the pool of adoptable children – and that includes survey content directly relevant to this population, such as adoption intentions and motivations/deterrents.

CONCLUSION

Nonparental caregivers may be more likely to consider adoption or take action to adopt the children in their care when the caregiver is a nonkin foster parent, a foster care agency or CPS was involved in the placement of the child, and/or financial assistance is available. They may be less likely to adopt if the child is older (especially among teenagers), if the caregiver is in poor health or is married, or if the caregiver type is kin foster care. Those with current plans to adopt but who have not yet adopted may face barriers to adoption such as extreme poverty, lower education and being unmarried. Discouraging reasons given for not considering or planning adoption tend to reflect issues with the birth parents more than issues with the child. However, the majority of nonadopting caregivers did not endorse any of the reasons assessed in the survey, suggesting that there are many unique reasons why nonparental caregivers do not adopt the children in their care.

References

  • Associated Press. [Accessed 4/21/2015];Married couples get adoption preference. 2011 at http://www.fox10phoenix.com/story/18119923/married-couples-get-adoption-preference.
  • Berrick JD, Barth RP, Needell B. A comparison of kinship foster homes and foster family homes: Implications for kinship foster care as family preservation. Children and Youth Services Review. 1994;16(1/2):33–63.
  • Bramlett MD. When stepparents adopt: Demographic, health and health care characteristics of adopted children, stepchildren, and adopted stepchildren. Adoption Quarterly. 2010;13(3/4):248–267.
  • Bramlett MD, Radel LF, Blumberg SJ. The health and well-being of adopted children. Pediatrics. 2007;119(S1):S54–S60. [PubMed]
  • Bussiere A. The Development of Adoption Law. Adoption Quarterly. 1998;1(3):3–25.
  • Carnochan S, Moore M, Austin MJ. Achieving Timely Adoption. Journal of Evidence-Based Social Work. 2013;10:2010–219. [PubMed]
  • Case A, Paxson C. Mothers and others: Who invests in children’s health? Journal of Health Economics. 2001;20:301–328. [PubMed]
  • Centers for Disease Control and Prevention (CDC), National Center for Health Statistics, State and Local Area Integrated Telephone Survey. 2011–2012 National Survey of Children’s Health Frequently Asked Questions. 2013 Apr; Available from: http://www.cdc.gov/nchs/slaits/nsch.htm.
  • CDC, National Center for Health Statistics, State and Local Area Integrated Telephone Survey. National Survey of Children in Nonparental Care: Frequently Asked Questions and Guidelinesfor Data Users. 2014 Apr; Available from: http://www.cdc.gov/nchs/slaits/nscnc.htm.
  • Connell CM, Katz KH, Saunders L, Tebes JK. Leaving foster care – The influence of child and case characteristics on exit rates. Children and Youth Services Review. 2006;28:780–798.
  • Gardino SL, Russell AE, Woodruff TK. Adoption After Cancer: Adoption Agency Attitudes and Perspectives on the Potential to Parent Post-Cancer. Cancer Treatment Research. 2010;156:153–170. [PMC free article] [PubMed]
  • Harris Interactive and the Dave Thomas Foundation for Adoption. National foster care adoption attitudes survey. 2007 Retrieved from www.davethomasfoundation.org/getdoc/9999a5f5-33ed-40fb-aedc-a676adc98546/Harris_DTFA-Report_FINAL_2_29_08.
  • Jones J. Vital and Health Statistics. 27. Vol. 23. Hyattsville, MD: National Center for Health Statistics, U.S. Department of Health and Human Services; 2008. Adoption experiences of women and men and demand for children to adopt by women 18–44 years of age in the United States, 2002. Retrieved from www.cdc.gov/nchs/data/series/sr_23/sr23_027.pdf. [PubMed]
  • Kirk HD. Shared Fate: A Theory and Method of Adoptive Relationships. Port-Angeles, WA: Ben-Simon; 1984.
  • Leathers SJ, Spielfogel JE, Gleeson JP, Rolock N. Behavior problems, foster home integration and evidence-based behavioral interventions: What predicts adoption of foster children? Children and Youth Services Review. 2012;34:891–899. [PMC free article] [PubMed]
  • Malm K, Vandivere S, McKlindon A. Children Adopted from Foster Care: Child and Family Characteristics, Adoption Motivation, and Well-Being. Washington: U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation; 2011.
  • McDonald T, Press A, Billings P, Moore T. Partitioning of the adoption process to better predict permanency. Child Welfare. 2007;86(3):5–32. [PubMed]
  • Merritt, Darcey Placement preferences among children living in foster or kinship care: A cluster analysis. Children and Youth Services Review. 2008;30:1336–1344.
  • Radel LF, Bramlett MD, Waters A. Legal and informal adoption by relatives in the U.S.: Comparative characteristics and well-being from a nationally representative sample. Adoption Quarterly. 2010;13(3/4):268–291.
  • Research Triangle Institute. SUDAAN Language Manual, Release 10.0. Research Triangle Park, NC: Research Triangle Institute; 2008.
  • Snowden J, Leon S, Sieracki J. Predictors of children in foster care being adopted: A classification tree analysis. Children and Youth Services Review. 2008;30:1318–1327.
  • Stoltzfus E. Child Welfare: The Fostering Connections to Success and Increasing Adoptions Act of 2008 (CRS Report RL34704) Washington, DC: The Congressional Research Service; 2008. [Accessed 4/21/2015]. at http://assets.opencrs.com/rpts/RL34704_20081009.pdf.
  • U.S. Department of Health and Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children's Bureau. [Accessed 4/9/2015];The AFCARS Report, No. 21: Preliminary Estimates for FY 2013 as of July 2014. 2014a at http://www.acf.hhs.gov/sites/default/files/cb/afcarsreport21.pdf.
  • U.S. Department of Health and Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children's Bureau. [Accessed 4/9/2015];Trends in Foster Care and Adoption: FFY 2002-FFY 2013. 2014b at http://www.acf.hhs.gov/sites/default/files/cb/trends_fostercare_adoption2013.pdf.
  • Vandivere S, Malm K, Radel L. Adoption USA: A Chartbook Based on the 2007 National Survey of Adoptive Parents. Washington, D.C: The U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation; 2009. [Accessed 4/15/2015]. at http://aspe.hhs.gov/hsp/09/NSAP/chartbook/
  • Wilson RF, Wilcox WB. Bringing Up Baby: Adoption, Marriage, and the Best Interests of the Child. William & Mary Bill of Rights Journal. 2006;14(3):883–908.
  • Visser SN, Zablotsky B, Holbrook JR, Danielson ML, Bitsko RH. National Health Statistics Reports. Vol. 81. Hyattsville, MD: National Center for Health Statistics, U.S. Department of Health and Human Services; 2015. Diagnostic experiences of children with Attention-deficit/Hyperactivity Disorder. Retrieved from http://www.cdc.gov/nchs/data/nhsr/nhsr081.pdf.
  • Yampolskaya S, Sharrock P, Armstrong MI, Strozier A, Swanke J. Profiles of children placed in out-of-home care: Association with permanency outcomes. Children and Youth Services Review. 2014;36:195–200.
  • Zill N, Bramlett MD. Health and well-being of children in foster care and children adopted from foster care. Children and Youth Services Review. 2014;40:29–40.