Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 May 1.
Published in final edited form as: Dev Psychobiol. 2020 Dec 10;63(5):1167–1176. doi: 10.1002/dev.22067

Parental Emotional Support and Social Buffering in Previously-Institutionalized and Typically Developing Children and Adolescents

Nicole B Perry 1, Anna E Johnson 3, Camelia E Hostinar 4, Megan R Gunnar 2
PMCID: PMC11062349  NIHMSID: NIHMS1969968  PMID: 33305356

Abstract

The current study tested the hypothesis that variation in parental emotional support explains differences in cortisol reactivity among 159 youth, including both previously-institutionalized (PI; N = 78) and non-adopted (NA; N = 81) children (ages 9-10) and adolescents (ages 15-16). Youth participated in a Modified Trier Social Stress Test after a period of preparation with either their parent or a supportive stranger. Saliva samples were collected to derive a measure of cortisol reactivity. Our findings revealed that parents buffered the cortisol stress response for PI children and adolescents only if they were high in emotional support. Our results also suggest that simply preparing with a parent might buffer the stress response for NA children; for NA adolescents, however, only emotionally supportive parents significantly buffered their adolescents’ stress.

Keywords: cortisol, social buffering, post-institutionalized

It is well known that the presence of attachment figures can block or reduce the activation of children’s stress physiology (Gunnar, Hostinar, Sanchez, Tottenham, & Sullivan, 2015). This effect is known in the literature as parental social buffering. Although previous work has provided valuable insight into parents’ ability to buffer the effects of stress for young, typically developing children (Gunnar & Donzella, 2002; Hennessy, Kaiser, & Sachser, 2009), whether and how parental social buffering changes as a result of early parental deprivation, child age, and specific parenting behaviors has been investigated less frequently. Studies published from our laboratory have been some of the first to address these important questions. In the current study, we build on our previous work by including previously-institutionalized (PI) children and adolescents, and by investigating the extent to which a specific parenting behavior (i.e. emotional support) plays a role in parents’ ability to serve as social buffers for previously-institutionalized (PI) and non-adopted (NA) youth.

Parental Social Buffering and Adverse Early-Life Experiences

Scientists have been interested in parental deprivation for decades (Bowlby, 1951; Rutter, 1972). Children adopted internationally into highly supportive families from institutions (e.g. orphanages) characterized by severe neglect and high caregiver turnover provide a unique opportunity to study the effects of these early adverse experiences. Early life is a critical time for developing secure caregiver attachments that build a child’s sense of trust and safety (Sroufe, Egeland, Carlson, & Collins, 2005). The sense of security that comes from knowing a trusted caregiver will be there to protect if needed allows children to explore and learn from their environment in ways that facilitate adaptive social and emotional functioning. Given that PI children experience neglect during what may be a sensitive period for the development of these important social bonds (Smyke, Dumitrescu, & Zeanah, 2002), it is not surprising that previous research finds PI children to have poorer social and emotional outcomes than their NA peers (Batki, 2018; Hawk & McCall, 2010), as well as suboptimal hormonal and neural responses to stress (Hodel et al., 2015; Tarullo & Gunnar, 2006).

Although prior work has shed considerable light on the effects of early deprivation for subsequent development, few studies have been able to address whether, years after adoption, PI youth are able to experience the same parental social buffering of stress as their NA counterparts. Previous work showing that PI children often fail to differentiate between parents and strangers at neural and physiological levels (Olsavsky et al., 2013; Wismer Fries, Shirtcliff, & Pollak, 2008) suggests that parents might not effectively buffer cortisol reactivity for PI youth. We only know of one study in addition to our own work that has examined this question. Wismer Fries and colleagues (Wismer Fries et al., 2008) found that 4-5-year-old children adopted from orphanages did not reduce cortisol when in physical contact with their mothers versus a stranger, whereas children who were not adopted showed these changes. In our work, we found that although parents were able to buffer the cortisol response to a social-evaluative stressor for 9-10-year-old NA children, parents were not effective social buffers for PI children’s cortisol responses in this context (Hostinar, Johnson, & Gunnar, 2015a).

The current study builds on our previous work by including both children and adolescents when examining whether parents can serve as effective social buffers, a question that has yet to be addressed during the adolescent period for PI youth. Although our prior findings with this sample did not indicate that parents buffer cortisol reactivity for PI children, it is possible that a longer history with adoptive parents, or changes in the hypothalamic pituitary adrenal (HPA) axis that occur throughout puberty, may play a role in parents’ ability to buffer cortisol reactivity for PI adolescents. For example, recent work from our laboratory provided empirical support for the recalibration of the HPA axis during puberty for PI youth such that increases in pubertal stage were associated with more normative increases in cortisol reactivity (Gunnar, DePasquale, Reid, & Donzella, 2019). Thus, if the HPA axis can recalibrate across puberty, parents’ ability to serve as a social buffer may change as well.

It should be noted, however, that in previous work with this sample we found that parents were able to buffer cortisol reactivity for NA children, but not NA adolescents (Hostinar, Johnson, & Gunnar, 2015b). Therefore, if HPA axis recalibration in PI adolescents makes their stress responses more directly comparable to those of their NA peers, parents may never serve as effective social buffers of the HPA axis for PI youth. Nonetheless, given that the transition to adolescence is characterized by increased stress, and PI youth have been found to experience greater levels of anxiety and depression as they progress through adolescence relative to their NA counterparts (Hawk & McCall, 2010), an empirical investigation of potential buffering effects for PI youth during adolescence is warranted.

Parental Social Buffering and Parenting Quality

The quality of the parent-child relationships is important in the development of the child’s capacity to regulate stress, even when the parent is not immediately available. For example, prior work has shown differential stress responses to social challenge by levels of parental structure (i.e. organization and consistency) in middle childhood (Ellenbogen & Hodgins, 2009), and measures of parental responsiveness in 4-year-olds (Hackman et al., 2013). Similarly, early parent-child relationships have also been found to be associated with adult stress reactivity (Pierrehumbert, Torrisi, Ansermet, Borghini & Halfon, 2012), though null results have also been reported (Ditzen et al., 2008).

Together, these studies highlight the importance of considering parenting quality when examining parental social buffering of children’s stress reactivity. However, the extent to which parenting quality matters for youth’s stress reactivity beyond childhood into adolescence and adulthood is less clear. To our knowledge, except for our previous work on this topic (Hostinar, Johnson, & Gunnar, 2015a), no study has considered parent-child relationship quality as it relates to social buffering in PI children and adolescents. In our prior work, we used broad measures of parental support as control variables and results suggested that parental support was not associated with our parental social buffering findings for PI or NA youth, a surprising finding given prior research in this area. This led to the question of whether specific parenting characteristics served as stronger predictors of parent’s ability to serve as social buffers than general measures of parenting quality. Because the stress paradigm used in the current study was designed to elicit emotions such as anxiety and fear, and PI youth have been reported to experience greater levels of anxiety (Hawk & McCall, 2010), parental emotional support may be a particularly important aspect of parenting that is salient for social buffering in PI youth. That is, although parents and children may report a generally positive relationship characterized by parental structure and responsiveness, how parents specifically respond to and help PI youth through emotions may have a greater impact on their ability to serve as a social buffer during times of emotional stress. Further, emotional support may be a particularly important aspect to consider when examining social buffering in typically developing adolescents. Even for NA youth, adolescence is a period of rapid growth such that there are hormonal and physical pubertal changes, but also more complex social relationships, increased stress, and higher expectations. It is during this period that youth begin to rely less on parents for social support. Thus, for parents to continue to buffer cortisol reactivity for NA adolescents, they may need to show greater emotional support. If true, this may explain why we previously found parents to serve as social buffers only among NA children and not NA adolescents.

The current study therefore extended prior work by investigating whether parents’ emotional support played a role in their ability to buffer cortisol reactivity for both NA and PI children and adolescents. Identifying specific parental behaviors that may influence the extent to which parents can buffer cortisol reactivity not only significantly adds to our knowledge in this area, but also has important implications for adoptive parents aiming to help youth cope with daily challenges.

The Current Study

In the current study, PI and NA children and adolescents prepared with either a parent or a supportive stranger prior to completing a social-evaluative stress task. Our primary question was whether parents’ level of emotional support and participant age moderated the association between preparation condition (preparing for a social stressor with a parent versus a stranger) and youth’s cortisol reactivity to the stressor. We hypothesized that preparing for a socially stressful task with a parent as opposed to a supportive stranger would buffer cortisol reactivity and that emotionally supportive parenting would be associated with the smallest increases in cortisol in response to a stress paradigm. Moreover, we hypothesized that emotionally supportive parenting may be particularly important for PI youth compared to NA youth. Finally, we hypothesized that emotional support may be particularly important for adolescents given that feeling supported by a caregiver during this time may require more than just being available or having a positive relationship.

Methods

Participants

Participants included 159 youth; 81 were typically developing youth born and raised in their biological families in a large urban Midwestern area (non-adopted, NA). In the typically developing group, half of participants were children (N = 40, M age = 9.97 years, SD = .52, range 9.1–11.1; 50% female) and half were adolescents (N = 41, M age = 16.05 years, SD = .39, range 15.2–16.8; 49% female). Parents reported a range of annual household income from under $35,000 for 3.7% of the sample to over $200,000 for 11.1%, with a median yearly family income of $100,000–$125,000. Only two parents did not report their household income. Parental education ranged from less than a high school degree to doctorate level, with a median of 16 years of education for the parents attending the session as well as their spouses or partners. NA participants were recruited from a department-maintained registry of families interested in research participation.

The sample also included 78 previously-institutionalized (PI) youth who were internationally adopted, half of whom were children (N = 38, M age = 9.72 years, SD = .57, range 8.9–11.0; 53% female) and half whom were adolescents (N = 40, M age = 15.57 years, SD = .67, range 15.2–16.8; 55% female). PI participants and their parents were recruited from the International Adoption Project registry at the University of Minnesota. The registry has been recruiting adopting families across the state for many years, drawing from a diverse set of adoption agencies. Youth spent an average of 21 months in institutional care (SD = 12.8, range 6-85). Adoptive parents reported a range of annual household income from under $25,000 for 2.7% of the sample to over $200,000 for 16%, with a median yearly family income of $75,000–$125,000. Only three parents did not report their household income. Similar to the NA group, parental education ranged from less than a high school degree to doctorate level, with a median of 16 years of education for the parents attending the session as well as their spouses or partners. The PI and NA youth did not differ significantly on any of the measured demographics.

The NA and PI data were concurrently collected. The sample size for each group was calculated a priori based on the large effect sizes observed in adult social buffering studies for the difference between peak cortisol responses in the stranger support versus romantic partner support conditions (e.g., Cohen’s d = .83 in Kirschbaum, Klauer, Filipp, & Hellhammer, 1995). Exclusion criteria for all participants were Autism Spectrum Disorder, Fetal Alcohol Syndrome, or any other major developmental disorder; and use of steroid medications (due to their interference with cortisol assay results). The majority of parents attending the session were mothers (88.8%), however there were no significant differences on any of the main study variables for children accompanied by their mothers versus their fathers, thus final analyses include both.

Two previous studies have reported on subsets of the sample used in the current study, but no study has used the full sample as is done here. The goal of one study was to examine whether parents served as social buffers for NA and PI youth in childhood (Hostinar, Johnson, & Gunanr, 2015a). Therefore, only PI and NA children (ages 9-10) were included. In the other study, the goal was to examine whether parents’ ability to serve as social buffers for NA youth changed from childhood to adolescence (Hostinar, Johnson, & Gunnar, 2015b). Thus, only NA children and adolescents were included. The current study extends this work by including PI adolescents and is the first to investigate emotion coaching as a specific aspect of parental support that may be particularly salient for social buffering.

Procedures

Stress paradigm.

The following research protocol was approved by the Institutional Review Board at the University of Minnesota. Participants completed an adapted version of the TSST using the instructions from the Modified Trier Social Stress Test (TSST-M, Yim et al., 2010; an adaptation of the TSST-C by Buske-Kirschbaum et al., 1997). This paradigm consisted of a public speaking task (introducing oneself to a hypothetical new classroom of students) and a mental arithmetic task (subtracting out loud by 7s from 758 for adolescents or by 3s from 307 for children). The participant was alone in the room when giving the speech and performing the mental arithmetic in front of a two-way mirror and a conspicuously placed video camera. The participant was told that the experimenter and two other teachers (one man, one woman) would watch them from the other side of the mirror and rate their speech performance and their arithmetic accuracy. This was accomplished using an audio recording of two adults who provided instructions for the speech as if they were present behind the two-way mirror. Replacing an audience of three judges with a two-way mirror has been previously shown to be successful in elevating cortisol in 9-year-olds (Jansen et al., 2000). If at any time the participant stopped talking during the speech portion of the task or made a mistake during the mental arithmetic, audio feedback was given through a microphone connected behind the double-sided mirror. At the end of the session, all participants and parents were debriefed about the protocol and given positive feedback on their performance.

Session timeline.

Participants were accompanied by one of their parents to a laboratory session with a start time scheduled between 3:30–4:30 p.m. to control for the diurnal variation in cortisol levels. The session included the following: consent process and reading leisurely in the welcome/waiting area (25 min.), participant moving to adjacent room with either the parent or a female experimenter (based on random assignment to the parent or stranger condition) and receiving TSST instructions (5 min.), speech preparation with parent or stranger (5 min.), moving to another room to complete the TSST alone (10 min.), returning to the waiting area and relaxing with the parent (10 min.), and completing questionnaires (approximately 1 hr). In the parent support condition, parents were instructed to help their child in any way they found useful. In the stranger support condition, the female stranger stated that she was ready to help in any way participants found useful. In this latter condition, the parent remained in the waiting room and did not accompany the participant into the testing area. Participants from both conditions could not see or contact their parents during the speech and math portion of the test. Salivary cortisol was collected four times (45, 65, 85, and 105 min after arrival), corresponding to 20 min after the end of the relaxation period and 20, 40, and 60 min after the end of the stress task.

Measures

Salivary cortisol.

Participants expelled saliva through a straw into pre-labeled vials. They were instructed to refrain from eating large, protein-filled meals and consuming caffeine or energy drinks 2 hours before arriving to the laboratory. The samples were stored in a laboratory freezer at −20 °C until being shipped to the University of Trier, Germany to be assayed using a time-resolved fluorescence immunoassay (dissociation-enhanced lanthanide fluorescent immunoassay [DELFIA]; intra-assay coefficient of variation <7%, interassay coefficient of variation <10%). All of the samples from each participant were included in the same assay batch, and the assay batches were balanced by group and condition. Samples were assayed in duplicate and averaged. All sample collections were timed and prompted by experimenters, thus, missing data rates were low (N = 5; .006% of samples). Cortisol values were skewed and therefore were transformed using natural log. Cortisol reactivity was derived by calculating a change score between the first sample taken (20 min after the end of the relaxation period) and the second sample taken (20 min following the end of the stress task). This a priori analytic choice was based on a previous meta-analysis suggesting that cortisol peaks 20-30 minutes post-stressor (Dickerson & Kemeny, 2004).

Daily diaries.

The parent and the child each completed a daily diary on session days, for which they reported information about the participant relevant to cortisol collection: time of wake-up, medication usage, caffeine consumption, distressing events experienced that day (e.g., arguments with siblings or parents), and number of hours of sleep during the previous night. Child reports were the primary source of information. However, for type of medication used by the offspring, the parents’ report on this variable was used instead when the child’s information was missing, too vague, or incomplete. All other child-reported variables were either complete or missing less than 5% of data so imputation was not necessary (Alice, 2015; Jakobsen, Gluud, Wetterslev, & Winkel, 2017). These variables were considered as potential covariates and included when significantly correlated with the outcome.

Emotional support.

Emotional support was measured via the Maternal Emotional Styles Questionnaire (MESQ; Lagacé-Séguin & Coplan, 2005) completed by the parent who accompanied the youth during the assessment (Nmother =150; Nfather=9). This measure was adapted from the meta-emotion interview (MEI-revised; Katz & Gottman, 1999), and assessed emotion-coaching and emotion-dismissing perspectives. Because we were primarily interested in whether parents could serve as a social buffer if they were supportive, only the emotion-coaching subscale was used in the current study. The form consisted of 14 statements that described parental attitudes and beliefs about their children’s emotions and about child emotion in general. Parents were asked to rate the degree to which they agreed or disagreed with each statement on a scale ranging from 1(strongly disagree) to 5 (strongly agree). Examples of the seven emotion-coaching statements (summed for the coaching composite) included, “When my child is sad, it’s an opportunity for getting close” and “When my child is angry, I want to know what he/she is thinking”. The MESQ is internally consistent, highly correlated with the original MEI, and convergent validity with parental goals and maternal depression has been established (Lagacé-Séguin & Coplan, 2005; Lagacé-Séguin & d’Entremont, 2006). The emotion-coaching subscale had an internal reliability of α = .72 in the current study. Higher scores on the emotion-coaching subscale indicate greater emotional support.

Data Analytic Plan

Multiple regression analyses were conducted in Mplus (Version 8; Muthén & Muthén, 2017) and Full Information Maximum Likelihood (FIML) was used to handle missing data. Less than 5% of the data were missing overall. Time since waking, family income, child age, child sex, and stressful life events within the past 3 months were all considered as potential covariates. However, none of these variables were correlated with emotional support or cortisol reactivity and were therefore excluded from final analyses. Use of medication was correlated with cortisol reactivity and was therefore included as a covariate in all analyses. In addition, because our measure of cortisol reactivity was obtained by calculating the change in cortisol from immediately post-TSST to peak cortisol after the stress paradigm, and change scores can be affected by where participants start, initial cortisol (time 1) was also entered as a covariate in all analyses. Finally, for PI participants, duration of institutional care prior to adoption was included as a covariate.

Unfortunately, we did not have a large enough sample to test a four-way interaction (group by condition by emotional support by age). Thus, the appropriate covariates, child age (child = 0, adolescent = 1), condition (parent = 0, stranger = 1), parental emotional support, the two-way interactions (emotional support by age, emotional support by condition, age by condition), and the three-way interaction (condition by emotional support by age) were regressed on to cortisol reactivity in separate models for the PI and NA participants. Models were then reduced accordingly based on non-significant findings.

Results

Preliminary Analyses

Preliminary analyses were conducted to examine descriptive information for study variables by group, condition, and age (see Table 1). For children, mean differences in emotional support (t(76) = .98, p = .33) or cortisol reactivity (t(76) = 1.25, p = .22) did not emerge by group (PI versus NA). We also did not see mean differences in emotional support (t(76) = −.77, p = .44) or cortisol reactivity (t(76) = .50, p = .62) by condition (parent vs. stranger) for children. For adolescents, emotional support did not differ by group (t(79) = .93, p = .36) or condition (t(79) = −.93, p = .36); mean differences in cortisol reactivity also did not emerge when comparing across condition (t(79) = .08, p = .94). However, overall, cortisol reactivity during the stress paradigm was significantly higher for the NA adolescents compared to the PI adolescents (t(79) = 2.19, p = .03).

Table 1.

Descriptive Statistics of Study Variables by Age and Group for each Condition

Previously-Institutionalized
Parent
 Stranger
Study Variables N Mean SD Skew (SE) N M SD Skew (SE)
Children (ages 9-10)
   Emotional Support 20 3.57 .45 −.178 (.51) 18 3.59 .65 −1.74 (.54)
   Cortisol Reactivity 20 .00 .11 −1.25 (.51) 18 .00 .17 −1.66 (.54)
   *Baseline Cortisol 20 4.03 5.55 4.05 (.51) 18 5.01 6.12 3.09 (.54)
   *TSST-M Cortisol 20 4.07 3.70 2.77 (.51) 18 4.26 3.77 1.42 (.54)
Adolescents (ages 15-16)
   Emotional Support 21 3.39 .57 −.40 (.50) 19 3.34 .51 −.16 (.52)
   Cortisol Reactivity 21 .00 .11 .10 (.50) 19 .00 .07 −2.28 (.52)
   *Baseline Cortisol 21 4.73 5.40 3.77 (.50) 19 4.53 3.39 2.09 (.52)
   *TSST-M Cortisol 21 4.54 4.50 2.37 (.50) 19 4.59 2.69 1.88 (.52)
Non-Adopted
Parent
 Stranger
Study Variables N Mean SD Skew (SE) N M SD Skew (SE)
Children (ages 9-10)
   Emotional Support 20 3.80 .37 −.83 (.51) 20 3.59 .60  −.28(.51)
   Cortisol Reactivity 20 .00 .07 −.49 (.51) 20 .05 .13  1.45 (.51)
   *Baseline Cortisol 20 3.19 2.39 1.78 (.51) 20 3.31 2.52  1.75 (.51)
   *TSST-M Cortisol 20 3.08 2.21 1.48 (.51) 20 4.67 4.11  1.81 (.51)
Adolescents (ages 15-16)
   Emotional Support 21 3.54 .50 .17(.50) 20 3.39 .37   .05 (.51)
   Cortisol Reactivity 21 .05 .09 .48(.50) 20 .04 .13  1.82 (.51)
   *Baseline Cortisol 21 4.61 1.80 −.17(.50) 20 4.80 2.91  1.49 (.51)
   *TSST-M Cortisol 21 6.00 3.53 .39(.50) 20 6.02 4.21  1.35 (.51)
Open in a new tab

note: values reported are pre-transformation and are in nmol/L

Primary Analyses

First, we conducted the model for NA participants (see Table 2a). The three-way interaction of age by condition by emotional support was significant and therefore subsumed the lower-level two-way interaction and main effects. The significant three-way interaction indicated that the two-way interaction of condition by support differed by age group. Thus, to probe the two-way interaction for each age group, we ran separate models for NA children and adolescents. As can be seen in Table 2b, for NA children, the condition by emotional support interaction was not significant, indicating that the extent to which the parent was emotionally supportive had no effect on the association between condition and cortisol reactivity. Surprisingly, the main effect of condition was also not significant, though it was trending in the expected direction with cortisol reactivity being lower when NA children prepared with a parent. It is important to note here that the sample size for this analysis was small (N = 40). Previously, using an analysis with more degrees of freedom because fewer variables were considered, a significant condition effect was obtained and reported (Hostinar. Johnson, & Gunnar, 2015b).

Table 2.

Standardized regression coefficients predicting cortisol reactivity for NA youth

Model Covariates Estimate (SE) z-value
a. Overall Model (NA Children and Adolescents)
   Cortisol at time 1 −.15 (.10) −1.48
   Medication use  .21 (.10)  2.05*
   Age group (child = 0 adolescent =1)  .16 (.15)  1.06
   Condition (stranger = 0 parent = 1) −.22 (.15) −1.42
   Emotional support  .04 (.17)   .23
   Emotional support x age group  .44(.21)  2.10*
   Emotional support x condition −.06 (.21)  −.31
   Condition x age group  .04 (.19)   .23
   Emotional support x condition x age group −.47 (.23) −2.02*
b. NA Children
   Cortisol at time 1 −.25 (.15) −1.72
   Medication use −.12 (.15)  −.75
   Condition (stranger = 0 parent = 1) −.26 (.16) −1.67ϯ
   Emotional support −.02 (.18)   −.10
   Emotional support x condition −.02(.19)   −.12
c. NA Adolescents
   Cortisol at time 1 −.09 (.13)   −.71
   Medication use .37 (.13)  2.99*
   Condition (stranger = 0 parent = 1) −.17 (.14) −1.25
   Emotional support  .58 (.22)  2.60*
   Emotional support x condition −.67(.21) −3.16*
Open in a new tab

Note.

*

p < .05,

ϯ

p < .10

For NA adolescents, the condition by emotional support interaction was significant (see Table 2c), indicating the association between condition and cortisol reactivity differed depending on how emotionally supportive parents reported that they were. A regions of significance plot (Preacher, Curran, & Bauer, 2006) shows how the association between condition and cortisol reactivity changes across the entire range of parental emotional support. Results revealed that for NA adolescents, at low levels of emotional support (1.1 SD below the mean and lower; 17% of the sample) those who prepared with a parent had higher cortisol reactivity than those who prepared with a supportive stranger (see Figure 1), but when emotional support was just above average (.3 SD above the mean and higher; 22%), adolescents who prepared with a parent had lower cortisol reactivity than those who prepared with a stranger.

Figure 1.

Figure 1.

Open in a new tab

Regions of significance plot probing the interaction between condition and emotional support predicting cortisol reactivity during the TSST-M for NA adolescents

Note: The X axis represents parental emotional support (standardized). The Y axis represents the difference in expected cortisol for an individual in the parent condition (coded 1) versus the stranger condition (coded 0) controlling for all other model variables. The curved lines represent 95% confidence bands around the adjusted effect of condition on cortisol reactivity. When the solid line is below zero, youth who prepared with a stranger had greater cortisol than youth who prepared with a parent. When looking where the confidence bands do not include zero, we see that the effect of condition on cortisol reactivity is significant at 1.1 SD below the mean of support (N = 7; 17%) and at .3 SD above the mean of support (N = 9; 22%) for NA adolescents. Thus, at low levels of support NA adolescents who prepared with a parent are expected to have higher cortisol than NA adolescents who prepared with a supportive stranger. However, when support is just above the mean, NA adolescents who prepared with a parent are expected to have lower cortisol than children who prepared with a supportive stranger.

Next, we conducted the model for PI participants (see Table 3). The three-way interaction of age by condition by emotional support was not significant. The condition by age interaction and the emotional support by age interaction were also not significant. However, the interaction of condition by emotional support predicting cortisol reactivity was significant, suggesting this effect was similar across both children and adolescents. To probe this interaction, we removed the non-significant interaction terms from the model and ran the model again. The regions of significance plot revealed that for PI youth, at very low levels of emotional support (1.8 SD below the mean and lower; 8%) those who prepared with a parent had higher cortisol reactivity than those who prepared with a supportive stranger (see Figure 2). Similarly, at very high levels of emotional support (1.6 SD above the mean and higher; 8%), PI youth who prepared with a parent had lower cortisol reactivity than those who prepared with a stranger.

Table 3.

Standardized regression coefficients predicting cortisol reactivity for PI youth

Model Covariates Estimate (SE) z-value
Overall Model (PI Children and Adolescents)
   Duration of institutional care −.06 (.08) −0.74
   Cortisol at time 1 −.68 (.06) −11.36**
   Medication use −.08 (.08) −1.05
   Age group (child = 0 adolescent =1)  .08 (.08) 1.06
   Condition (stranger = 0 parent = 1)  .00 (.08) −0.05
   Emotional support  .12 (.11) 1.03
   Emotional support x condition −.25 (.11) −2.26*
Open in a new tab
*

p < .05

**

p < .001

Figure 2.

Figure 2.

Open in a new tab

Regions of significance plot probing the interaction between condition and emotional support predicting cortisol reactivity during the TSST-M for PI youth

Note: The X axis represents parental emotional support (standardized). The Y axis represents the difference in expected cortisol for an individual in the parent condition (coded 1) versus the stranger condition (coded 0) controlling for all other model variables. The curved lines represent 95% confidence bands around the adjusted effect of condition on cortisol reactivity. When the solid line is below zero, youth who prepared with a stranger had greater cortisol than youth who prepared with a parent. When looking where the confidence bands do not include zero, we see that the effect of condition on cortisol reactivity is significant at 1.8 SD below the mean of emotional support (N = 6; 8%) and at 1.6 SD above the mean of emotional support (N = 6; 8%). Thus, at very low levels of support PI youth who prepared with a parent are expected to have higher cortisol than PI youth who prepared with a supportive stranger. However, at very high levels of support, PI youth who prepared with a parent are expected to have lower cortisol than PI youth who prepared with a supportive stranger.

Discussion

One question that has received relatively little attention is whether children adopted after experiencing early neglect and deprivation are later able to effectively use caregivers as buffers against stress. In our previous work using this sample, we investigated this question in childhood (ages 9-10 years) and found that parents were social buffers for NA children but not PI children. Although social buffering effects were not present in childhood for PI youth, it is possible that a longer history with caregivers or pubertal changes of the HPA axis (Gunnar et al., 2019) allow parents to emerge as stress buffers for PI adolescents. To our knowledge, this is the first study to investigate this question.

To build on our previous work, we included both PI children and adolescents in the current analyses and tested emotional support as a specific aspect of the parent-child relationship. We first tested a three-way interaction to examine whether the association between emotional support and condition (preparation with parent versus preparation with stranger) was associated with cortisol reactivity differently for PI children and PI adolescents. Our findings revealed that the association between preparation condition and cortisol reactivity varied by level of emotional support, but this interaction worked similarly across both age groups. Our prior work indicated that including broader aspects of the parent-child relationship had no effect on parents’ inability to serve as a social buffer for PI children. Findings from the current study extend this work and suggest that parents may be able to buffer cortisol reactivity for PI children and adolescents if they typically provide extremely high levels of emotional support. Being a “super parent” with regard to emotional availability and emotional coaching may be what is needed for children who have difficulty developing a sense of security and trust following institutional care. This extra support may provide necessary reassurance that the caregiver is a reliable source of support. Indeed, research has reported a significant improvement in attachment quality following adoption placement (van IJzendoorn & Juffer, 2006), with a large contribution attributed to adoptive parents’ emotional availability (Barone, Lionetti, & Green, 2017).

These findings have significant implications for adoptive families such that parental emotion coaching and availability may be a target for intervention. Intervention work has already shown to be effective in this area. For example, Barone and colleagues (Barone, Barone, Dellagiulia, & Lionetti, 2018) found that mothers and children participating in an intervention aimed at increasing positive parenting were significantly higher in their emotional availability and sensitivity than mothers and children who did not participate. Children participating in the intervention also increased in their emotional availability, possibly making them more receptive to their mother’s emotional support and creating a positive developmental cycle.

When testing these associations among the NA participants, the three-way interaction was significant and indicated that the two-way interaction of condition (preparing with a parent versus stranger) by support differed for NA adolescents and children. For NA children, there was a trend-level association between condition and cortisol reactivity such that preparing with a parent was associated with lower cortisol reactivity. Previous work with this sample of children with increased power showed a significant main effect of condition when modeling cortisol as a growth curve (Hostinar, Johnson, & Gunnar, 2015b). Because we were interested in testing three-way interactions across multiple groups and identifying the level at which emotional support mattered for the association between condition and cortisol reactivity, we did not have a large enough sample to create growth curves of cortisol reactivity in the current study. Thus, it is likely that our small sample size limited our ability to detect this main effect. However, in combination with our previous work, these findings suggest that simply having a parent present is enough to buffer cortisol reactivity for NA children. It may be that in childhood, when children are still mostly reliant on caregivers for basic care and support, having a parent close by provides enough reassurance and security to buffer children’s stress response.

In contrast, in our previous work with this sample, we found that parents were not effective social buffers for NA adolescents even after considering broad measures of parenting (Hostinar, Johnson, & Gunnar, 2015b). The results of the present study qualify that finding. Specifically, the non-significant effect of condition for the NA adolescents was qualified by finding that at low levels of parent emotional support, preparing with the parent was actually associated with higher cortisol reactivity than preparing with a stranger. However, if the parents were just above average in reports of typical emotional support, they were effective social buffers of their adolescent’s cortisol reaction to the TSST. In combination with our prior work, these findings suggest that broader measures of parenting may not predict social buffering efficacy in adolescents, but parents can in fact serve as effective social buffers if they habitually engage in specific emotion-based parenting behaviors.

These findings have implications for the role of parents in reducing cortisol reactivity among typically developing adolescents. As children enter the adolescent period, they often rely on parents less frequently for social support. Thus, simply having a parent present may not be enough for adolescents to experience decreased stress reactivity. However, if parents are a source of emotional support through coaching of emotions and helping adolescents manage their emotional experiences, then being present during stress may allow them to serve as effective social buffers for NA youth.

The current study had several strengths and a number of limitations. Studying PI children allows us to better understand the negative effects of early adversity while at the same time examining how positive social interventions may weaken these effects. By including two age groups, we were able to test the role of parents as social buffers for PI youth at two developmental stages. For both the early adversity and typically developing groups, we were able to better understand how emotional support may uniquely affect parents’ ability to dampen youth’s stress responses by focusing on a very specific aspect of parenting. Finally, the use of an experimental stressor was advantageous in that it created the same stressful context for all youth and the experimental control of social support condition clarified correlational relationships between aspects of parenting and stress responses.

In addition to these strengths, there were several limitations. First, we did not have an observed measure of parental emotional support and had to rely on parents’ reports of their emotional supportiveness. An observational measure of general parental support was available but only for youth who prepared with their parent. This measure was correlated with parents’ reports of emotional support, although modestly (r = .15). Moreover, we did not have a measure of youths’ perceived helpfulness from their support person, which would have added valuable insight regarding youth’s awareness of the effectiveness of the support received. Future work observing parents’ emotional support and asking youth to report on the extent to which they felt supported by their parent or a stranger would shed additional light on these associations.

Second, we did not have a large enough sample to model cortisol in a more detailed way (e.g., slopes) while still testing multiple interactions. Instead, cortisol reactivity was derived by calculating a change score between the first sample taken (20 min after the end of the relaxation period) and the participant’s peak cortisol level after the TSST-M. We chose this measure because it adequately reflects the peak cortisol response without including cortisol levels during recovery. Future studies with a larger sample should replicate these results. Finally, the cross-sectional design precludes conclusions regarding developmental change. For example, it would be beneficial to know whether increases in parental emotional support are associated with decreases in children’s cortisol reactivity to stress over time. Future replication of these same questions longitudinally is warranted.

Despite these limitations, the present study adds important insight into the role of parental emotional support in social buffering. These results suggest that high level of emotional support, particularly for PI youth, is an important factor in the social regulation of HPA stress reactivity.

Data availability statement:

Research data are not shared

References

  1. Alice M (2015). Imputing missing data with R; MICE package. Available at https://datascienceplus.com/imputing-missing-data-with-r-mice-package/.
  2. Barone L, Barone V, Dellagiulia A, & Lionetti F (2018). Testing an attachment-based parenting intervention-VIPP-FC/A in adoptive families with post-institutionalized children: Do maternal sensitivity and genetic markers count? Frontiers in Psychology, 19, 156–177. 10.3389/fpsyg.2018.00156 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Barone L, Lionetti F, & Green J (2017). A matter of attachment? How adoptive parents foster post-institutionalized children’s social and emotional adjustment. Attachment and Human Development, 19(4), p. 323–339. 10.1080/14616734.2017.1306714 [DOI] [PubMed] [Google Scholar]
  4. Batki A (2018). The impact of early institutional care on emotion regulation: studying the play narratives of post-institutionalized and early adopted children. Early Child Development and Care, 188(12), 1005–1815. 10.1080/03004430.2017.1289190 [DOI] [Google Scholar]
  5. Bowlby J (1951). Maternal care and mental health. Bulletin of the World Health Organization, 3, 355–533. [PMC free article] [PubMed] [Google Scholar]
  6. Buske-Kirschbaum A, Jobst S, Wustmans A, Kirschbaum C, Rauh W, & Hellhammer D (1997). Attenuated free cortisol response to psychosocial stress in children with atopic dermatitis. Psychosomatic Medicine, 59(4), 419–426. http://www.ncbi.nlm.nih.gov/pubmed/9251162 [DOI] [PubMed] [Google Scholar]
  7. Dickerson SS, & Kemeny ME (2004). Acute stressors and cortisol responses: [DOI] [PubMed] [Google Scholar]
  8. A theoretical integration and synthesis of laboratory research. Psychological Bulletin, 130, 355–391. doi: 10.1037/0033-2909.130.3.355 [DOI] [PubMed] [Google Scholar]
  9. Ditzen B, Schmidt S, Strauss B, Nater UM, Ehlert U, & Heinrichs M (2008). Adult attachment and social support interact to reduce psychological but not cortisol responses to stress. Journal of Psychosomatic Research, 64, 479–486. doi: 10.1016/j.jpsychores.2007.11.011 [DOI] [PubMed] [Google Scholar]
  10. Ellenbogen MA, & Hodgins S (2009). Structure provided by parents in middle childhood predicts cortisol reactivity in adolescence among the offspring of parents with bipolar disorder and controls. Psychoneuroendocrinology, 34(5), 773–785. doi: 10.1016/j.psyneuen.2008.12.011 [DOI] [PubMed] [Google Scholar]
  11. Gunnar MR, DePasquale CE, Reid BM, and Donzella B (2019). Pubertal stress recalibration reverses the effects of early life stress in post-institutionalized children. Proceedings of the National Academy of Sciences, 116 (48), 23984–23988. 10.1073/pnas.1909699116 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gunnar MR, & Donzella B (2002). Social regulation of the cortisol levels in early human development. Psychoneuroendocrinology, 27,199–220. 10.1016/S0306-4530(01)00045-2 [DOI] [PubMed] [Google Scholar]
  13. Gunnar MR, Hostinar CE, Sanchez MM, Tottenham N, & Sullivan RM (2015). Parental buffering of fear and stress neurobiology: Reviewing parallels across rodent, monkey, and human models. Social Neuroscience 10(5), 474.478. 10.1080/17470919.2015.1070198 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hackman DA, Betancourt LM, Brodsky NL, Kobrin L, Hurt H, & Farah MJ (2013). Selective impact of early parental responsivity on adolescent stress reactivity. PLoS One, 8 (3), e58250. doi: 10.1371/journal.pone.0058250 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hawk B, & McCall RB (2010). CBCL behavior problems of post-institutionalized international adoptees. Clinical Child and Family Psychology Review, 13(2), p/ 199–211. 10.1007/s10567-010-0068-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hennessy MB, Kaiser S, & Sachser N (2009). Social buffering of the stress response: Diversity, mechanisms, and functions. Frontiers in Neuroendocrinology, 30, 470 – 482. 10.1016/j.yfrne.2009.06.001 [DOI] [PubMed] [Google Scholar]
  17. Hodel AS, Ruskin HH, Cowell RA, Van Den Heuvel SE, Gunnar MR, & Thomas KM (2015). Duration of early adversity and structural brain development in post-institutionalized adolescents. NeuroImage, 105(15), p. 112–119 10.1016/j.neuroimage.2014.10.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hostinar CE, Johnson AE, & Gunnar MR (2015a). Early social deprivation and the social buffering of cortisol stress responses in late childhood: An experimental study. Developmental Psychology, 51(11), p. 1597–1608. 10.1037/dev0000029 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hostinar CE, Johnson AE, & Gunnar MR (2015b). Parent support is less effective in buffering cortisol stress reactivity for adolescents compared to children. Developmental Science, 18(2), p. 281–297. DOI: 10.1111/desc.12195 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jakobsen JC, Gluud C, Wetterslev, & Winkel P (2017). When and how should multiple imputation be used for handling missing data in randomised clinical trials – a practical guide with flowcharts. BMC Med Res Methodol 17, 162. 10.1186/s12874-017-0442-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Jansen LMC, Gispen-de Wied CC, Van der Gaag RJ, ten Hove F, Willemsen- Swinkels SWM, Harteveld E, & van Engeland H (2000). Unresponsiveness to psychosocial stress in a subgroup of autistic-like children, multiple complex developmental disorder. Psychoneuroendocrinology, 25, 753–764. [DOI] [PubMed] [Google Scholar]
  22. Kirschbaum C, Klauer T, Filipp SH, & Hellhammer DH (1995). Sex-specific effects of social support on cortisol and subjective responses to acute psychological stress. Psychosomatic Medicine, 57, 23–31. http://www.psychosomaticmedicine.org/content/57/1/23.long [DOI] [PubMed] [Google Scholar]
  23. Lagacé-Séguin DG, & Coplan RJ (2005). Maternal emotional styles and child social adjustment: Assessment, correlates, outcomes, and goodness of fit in early childhood. Social Development, 14(4), p. 613–636. 10.1111/j.1467-9507.2005.00320.x [DOI] [Google Scholar]
  24. Lagacé-Séguin DG, & d’Entremont. (2006). The role of child negative affect in the relations between parenting styles and play. Early Child Development and Care, 176, p. 461–477. 10.1080/03004430500039960 [DOI] [Google Scholar]
  25. Muthén LK, & Muthén BO (2011). Mplus User’s guide, 6thed. Muthén & Muthén, Los Angeles: CA. [Google Scholar]
  26. Olsavsky AK, Telzer EH, Shapiro M, Humphreys KL, Flannery., Goff B, & Tottenham N (2013). Indiscriminate amygdala response to mothers and strangers after early maternal deprivation. Biological Psychiatry, 74, 853–860. 10.1016/j.biopsych.2013.0.025 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Pierrehumbert B, Torrisi R, Ansermet F, Borghini A, & Halfon O (2012). Adult attachment representations predict cortisol and oxytocin responses to stress. Attachment & Human Development, 14 (5), 453–476. doi: 10.1080/14616734.2012.706394 [DOI] [PubMed] [Google Scholar]
  28. Preacher KJ, Curran PJ, & Bauer DJ (2006). Computational tools for probing interactions in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31(4), p. 437–448. 10.3102/10769986031004437. [DOI] [Google Scholar]
  29. Rutter M (1972). Maternal deprivation reconsidered. Journal of Psychosomatic Research, 16, 241–250. 10.1016/0022-3999(72)90005-0 [DOI] [PubMed] [Google Scholar]
  30. Smyke AT, Dumitrescu A, & Zeanah CH (2002). Attachment disturbances in young children. I: The continuum of caretaking casualty. Journal of the American Academy of Child & Adolescent Psychiatry, 41,972–982. 10.1097/00004583-200208000-00016 [DOI] [PubMed] [Google Scholar]
  31. Sroufe LA, Egeland B, Carlson E, & Collins WA (2005a). The development of the person: The Minnesota study of risk and adaptation from birth to adulthood. New York: Guilford. [Google Scholar]
  32. Tarullo AR, & Gunnar MR (2006). Child maltreatment and the developing HPA axis. Hormones and Behavior, 50 (4), p. 632–639. 10.1016/j.yhbeh.2006.06.010 [DOI] [PubMed] [Google Scholar]
  33. Van Ijzendoorn MH, & Juffer F (2006). The Emanuel Miller Memorial Lecture 2006: Adoption as intervention. Meta-analytic evidence for massive catch-up and plasticity in physical, socio-emotional, and cognitive development. Journal of Child Psychology and Psychiatry, 47(12), p. 1228–1245. 10.1111/j.1469-7610.2006.01675.x [DOI] [PubMed] [Google Scholar]
  34. Wismer Fries AB, Shirtcliff EA, & Pollak SD (2008). Neuroendocrine dysregulation following early social deprivation in children. Developmental Psychobiology, 50, 588–599. 10.1002/dev.20319 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Yim IS, Quas JA, Cahill L, & Hayakawa CM (2010). Children’s and adults’ salivary cortisol responses to an identical psychosocial laboratory stressor. Psychoneuroendocrinology, 35(2), 241–248. doi: 10.1016/j.psyneuen.2009.06.014 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Research data are not shared

RESOURCES