© 2014 American Psychological Association 0893-3200/14/$ 12.00 http://dx.doi .org/10.1037/a0037328

Journal of Family Psychology 2014, Vol. 28, No. 4, 538-548

Parent-Child Relationship Quality Moderates the Link Between Marital Conflict and Adolescents’ Physiological Responses to Social Evaluative Threat Rachel G. Lucas-Thompson

Douglas A. Granger

Colorado State University

Arizona State University and Johns Hopkins University

This study examined how marital conflict and parent-child relationship quality moderate individual differences in adolescents’ adrenocortical and autonomic nervous system (ANS) responses to social evaluative threat. Saliva samples (later assayed for cortisol and alpha-amylase, sAA) were collected from 153 youth (52% female; ages 10-17 years) before and after, and cardiovascular activity was assessed before and during, the Trier Social Stress Test (TSST). Marital conflict predicted higher levels of sAA reactivity but lower levels of heart rate (HR) reactivity. Parent-child relationship quality moderated these associations, such that marital conflict was more strongly related to heightened sAA and dampened SBP reactivity if youth had low-quality relationships with their parents. The findings suggest a “dual-hazard” pathway with implications for biosocial models of the family, as well as theories of the social determinants of biological sensitivity/susceptibility to context. Keywords: marital conflict, parent-child relationships, cortisol, alpha-amylase, stress

echolamines into circulation (Chrousos & Gold, 1992). The ANS has two coordinated systems that often operate in opposition to each other; the SNS and the parasympathetic nervous system (PNS). These two branches of the ANS work interactively together to produce various levels of physiological arousal, and changes in cardiovascular parameters are the result of complex and often antagonistic interactions between the SNS and PNS (i.e., SNS activation accelerates the heart and constricts blood vessels, whereas PNS activation decelerates the heart (Porges, 1992). The HPA axis is particularly responsive to socially evaluative threat, and is considered a “slower” responding system than the ANS (Dickerson & Kemeny, 2004), whereas the ANS is responsible for activating “fight or flight” mechanisms; the ANS broadly and cardiovascular functioning specifically are considered “faster” re­ sponding than the HPA axis (Berntson, Cacioppo, & Quigley, 1991). Short-term activation of these systems is adaptive, but repeated and/or prolonged activation interferes with the body’s ability to stabilize after activation and therefore interferes with effective stress management. Chronic activation of the HPA and/or ANS is thought to translate into cumulative wear on physiological sys­ tems, a burden termed “allostatic load” (Juster, McEwen, & Lupien, 2010). In developmental science, theorists call for a multi­ system measurement strategy to index individual differences in the psychobiology of the stress response; in the current study, we included cortisol as a marker of HPA axis functioning, as well as salivary a-amylase (sAA) as a marker of ANS activation, and particularly SNS functioning (Granger, Kivlighan, el-Sheikh, Gordis, & Stroud, 2007). In addition, we included blood pressure (BP) and heart rate (HR) as markers of cardiovascular activation, par­ ticularly because they have been widely used as indicators of physiological functioning in past research (e.g., Bauer, Quas, & Boyce, 2002) and also because cardiovascular functioning is

Poor-quality family relationships put children at-risk for prob­ lems in multiple domains (e.g., Rhoades, 2008). Contemporary theories implicate a biosocial mechanism (Booth, Carver, & Granger, 2000) in that the family context moderates children’s capacities to coordinate biobehavioral responses to stress (Luecken & Lemery, 2004). Individual differences in children’s biological sensitivities are assumed to translate adversity into risk or resil­ ience (Belsky & Pluess, 2009; Boyce & Ellis, 2005). Historically, studies have focused on the isolated effects of either marital or parent-child relationships on children’s stress reactivity and reg­ ulation. The goals of this study were to determine how marital conflict and parent-child relationship quality, alone and in com­ bination, contribute to individual differences in adolescents’ biobe­ havioral responses to social evaluative threat.

Psychobiology of the Stress Response The primary systems involved in the psychobiology of stress are the hypothalamic-pituitary-adrenal (HPA) axis and the autonomic nervous system (ANS). Activation of the HPA axis results in the release of gluco-corticoids (i.e., cortisol); activation of the sympa­ thetic branch (SNS) of the ANS results in release of cat-

Rachel G. Lucas-Thompson, Human Development and Family Studies, Colorado State University; Douglas A. Granger, Institute for Interdisci­ plinary Salivary Bioscience Research, Arizona State University and School of Nursing and Bloomberg School of Public Health, Johns Hopkins Uni­ versity. Correspondence concerning this article should be addressed to Rachel G. Lucas-Thompson, 1570 Campus Delivery, Colorado State University, Fort Collins, CO 80523-1570. E-mail; Lucas-Thompson.Rachel.Graham@ colostate.edu 538

FAMILY RELATIONSHIPS AND PHYSIOLOGICAL REACTIVITY

strongly linked with health (e.g., Fried et al., 1998). Following past research (e.g., Ballard, Cummings, & Larkin, 1993) we included systolic (SBP; BP when the heart contracts/beats) and diastolic (DBP; BP between heartbeats) blood pressure. Several studies suggest that SBP may be more strongly related to interpersonal stressors than is DBP (e.g., Lucas-Thompson, 2012; Stroud et al., 2009).

Family Relationships, Stress, and the Stress Response Guiding the current study were theoretical frameworks that emphasize the importance of the family context as a moderator of the expression of hormone-behavior relationships and stress reg­ ulation (Booth et al., 2000; Luecken & Lemery, 2004), highlight the potential for multiple types of physiological dysregulation in the face of chronic stress (Laurent, Powers, & Granger, 2013), and posit that chronic stress likely leads to asymmetry in stress re­ sponding (i.e., distinct patterns of dysregulation across systems), and that such asymmetry is particularly damaging in terms of long-term health and behavioral outcomes (Bauer et al., 2002; Gordis, Granger, Susman, & Trickett, 2006). Participating in and/or observing poor-quality, high-conflict family relationships is stressful for children (Ballard et al., 1993; Davies, Sturge-Apple, Cicchetti, & Cummings, 2007; El-Sheikh, 1994). Although family relationships are not static, there is a great deal of stability in characteristics of these relationships (e.g., Waters, Merrick, Treboux, Crowell, & Albersheim, 2000). Poorquality, high-conflict family relationships are therefore a poten­ tially chronic stressor for children. Negative family relationships offer children few resources as well as poor models for effective coping (e.g., Chorpita & Barlow, 1998). Parents can serve as external regulators of children’s emotional and physiological stress responses (e.g., Kopp, 1989), but conflict often drains parents of the personal resources that support such buffering. Chronic stressors, like exposure to poor-quality family relation­ ships, can lead to various patterns of physiological dysregulation, including both heightened and blunted reactivity (Heim & Nemeroff, 1999). The hyperreactivity hypothesis (e.g., Cummings, Zahn-Waxler, & Radke-Yarrow, 1981) posits that chronic family stress leads to exaggerated physiological stress responses. In con­ trast, the hyporeactivity hypothesis suggests that chronic stress can lead to underreactive stress responses, especially with increasing exposure that eventually taxes individuals to the point that they can no longer respond (Kudielka, Hellhammer, & Wiist, 2009). At­ tachment insecurity predicts greater cortisol reactivity, particularly for temperamentally difficult children (e.g., Gunnar, Brodersen, Nachmias, Buss, & Rigatuso, 1996), whereas parental sensitivity predicts more rapid cortisol recovery from stress in children (e.g., Albers, Riksen-Walraven, Sweep, & de Weerth, 2008). A history of marital conflict exposure predicts sensitized SBP (Ballard et al., 1993) and HR (El-Sheikh, 1994) but dampened cortisol (Davies et al., 2007) reactivity to simulated marital conflict in children. Marital conflict predicts dampened cortisol and SBP reactivity to social-evaluative stress in late adolescence (Lucas-Thompson, 2012). Because very few studies have examined how family con­ flict is related to adolescents’ responses to social-evaluative stress, it is unclear whether this discrepancy about whether family con­ flict predicts dampened or sensitized SBP reactivity is driven by (a) developmental differences in the ways that family conflict

539

affects physiological responding over time (Lucas-Thompson, 2012); or (b) differences in cardiovascular responding to conflictspecific versus general stressors. Despite a plethora of studies linking family relationships to HPA axis activity and cardiovascular reactivity SBP and HR in children, few studies have examined how family relationships predict sAA regulation. An avoidant attachment predicts greater sAA reactivity (Hill-Soderlund et al., 2008), but father-adolescent relationship quality is related to cortisol but not sAA responses to a conflict discussion (Byrd-Craven, Auer, Granger, & Massey, 2012). More work is needed to examine how marital conflict and parenting uniquely and in interaction with each other predict children’s sAA responses to acute stress. The effects of stress are presumed to differ across developmen­ tal periods (Karatoreos & McEwen, 2013), depending on the malleability of the brain and body. Over time, responses to family conflict may become increasingly heightened (Cummings et al., 1981) or dampened because the stress response has been, in a sense, inoculated or overwhelmed by chronic stress activation (Kudielka et al., 2009). Adolescents experience a dramatic reor­ ganization of body and brain systems due to maturational change in neuroendocrine physiology, which is believed to result in in­ creased sensitivity to negative environmental stimuli (Andersen, 2003). Therefore, many have emphasized that it is particularly worthwhile to explore associations between family relationships and stress physiology during the adolescent transition (Stroud et al., 2009).

A Family Systems Approach Most studies that have examined the links between family relationships and physiological regulation have focused on isolated family relationships, despite theoretical arguments that it is impor­ tant to consider the effects of multiple and coordinated relation­ ships (Cox & Paley, 1997). Although maintaining high-quality relationships with children may be difficult in the face of marital conflict (e.g., Erel & Burman, 1995), doing so may buffer children from the negative effects of conflict on developmental trajectories (e.g., Grych & Fincham, 1990). It is also possible that negative characteristics of parent-child relationships may exacerbate the effects of marital conflict on child outcomes like stress reactivity. There is evidence that high-quality parenting buffers children from health and adjustment problems when there is marital conflict (e.g., Katz & Gottman, 1997), but other studies suggest main but not interactive effects of marital conflict and parenting (Grych, Raynor, & Fosco, 2004). The only studies to examine these pat­ terns in relation to stress physiology have examined cortisol, either diurnal or reactivity patterns. In adolescence, better marital func­ tioning and parenting independently predict lower cortisol levels and steeper diurnal cortisol rhythms (Pendry & Adam, 2007). During toddlerhood, maternal sensitivity buffers children exposed to partner violence from later developing sensitized cortisol reac­ tivity (Hibel, Granger, Blair, & Cox, 2011). Theory emphasizes that family relationships should be related to functioning across stress-response systems (Luecken & Lemery, 2004), but no studies have examined unique and joint contributions of marital and parent-child relationships to physiological indicators other than cortisol. No studies have examined these patterns in relation to stress reactivity beyond early childhood. To best understand long-

LUCAS-THOMPSON AND GRANGER

540

in the afternoon. After giving informed consent and assent, youth and parents were taken to separate rooms. Each family member separately filled out questionnaires using Audio Computer As­ sisted Self Interview software. Family members were each paid $20 for participation; an additional $10 covered transportation.

term risk, it is necessary to examine multiple arms of the stress response system (Granger et al., 2012), in part because chronic stress predicts distinct patterns of dysregulation across systems (Gordis, Granger, Susman, & Trickett, 2008).

The Current Study This study evaluated marital conflict and parent-child relation­ ships in relation to HPA axis (i.e., cortisol) and ANS (i.e., sAA and cardiovascular) responses to stress. In line with past research (Davies et al., 2007; Lucas-Thompson, 2012), we hypothesized that marital conflict would predict hyporeactive cortisol responses. Although there are contrasting findings about whether family conflict predicts hypo- or hyperreactive cardiovascular responses (e.g., El-Sheikh, 1994; Lucas-Thompson, 2012), there is more empirical support for the hyperreactivity hypothesis. Therefore, we anticipated that marital conflict would predict hyperreactive car­ diovascular responses. Because cardiovascular and sAA function­ ing are both indicators of ANS activity, we predicted similar patterns of dysregulation in these domains. Hypothesizing that patterns of dysregulation are distinct across stress response sys­ tems is in line with theory (Bauer et al., 2002) and evidence, particularly in adolescence (Gordis et al., 2006; Lucas-Thompson, 2012). We also expected that high-quality relationships with par­ ents would “protect” children from displaying dampened or sen­ sitized stress responses with greater marital conflict.

Method Participants One-hundred and fifty-three youth (52% female) between 10 and 17 years of age (M = 12.92, SD = 2.16) and both of their parents were recruited from the community through study adver­ tisements. Children came from 98 families; although most families were intact (78%), step-parent families were included if the adults had been married or cohabiting for at least 2 years (M = 15.64 years, SD = 5.86). Of those who reported ethnicity (6% did not), youth were 49% non-Hispanic Caucasian, 26% other or mixed ethnicities, 17% African American, 6% Asian American, 1% American Indian, and 1% Hispanic. Yearly family income ranged from $3,375 to $450,000 (Median = $67,750, SD = $63,879.39). On average, both mothers and fathers had completed an associate’s degree (or vocational training beyond high school).

Procedure: Overall Prior to the visit, youth were asked to avoid alcohol and cold medicine 24 hr before the visit and caffeine, eating, smoking, and exercise for 2 hours before the visit. Families visited the laboratory

Baseline

TSST

5( T

(

r

S Practice

SI

Figure 1.

Procedure: Youth A research assistant first attached a BP cuff to participants; then, the procedure for saliva collection was explained and practiced. Participants were instructed that they could put the swab anywhere except between gums and lips. They left swabs in for 3 min. Participants then filled out the shortened version of the Positive and Negative Affect Schedule (PANAS) (Laurent et al., 1999) and then sat quietly for 10 min, during which baseline BP and HR were recorded. Baseline saliva was then collected (see Figure 1 for a visit timeline). Theory emphasizes that family relationships have the potential to disrupt general physiological functioning (Luecken & Lemery, 2004). In addition, past research has provided evidence that poorquality family relationships disrupt stress responding to stressors that are unrelated to marital conflict or family stress (LucasThompson, 2012). Therefore, examined were responses to a mod­ ified Trier Social Stress Test (TSST) that is appropriate for chil­ dren and adolescents (Yim, Quas, Cahill, & Hayakawa, 2010b); this stressor was selected because it is a robust social-evaluative stressor that reliably produces physiological responses across sys­ tems (whereas other stressors often do not elicit responses across physiological systems; e.g., Kudielka & Kirschbaum, 2007). In this task, participants give a speech about personal characteristics and then engage in out-loud mental arithmetic. These activities are observed by a female evaluator who remains neutral, and who participants are told will examine their speech, posture, and tone of voice. The task is videotaped and children are told that the tape will be evaluated by experts. Immediately after the TSST, participants again completed the PANAS (Laurent et al., 1999). For the next 30 min, participants filled out questionnaires that were unlikely to affect physiological recovery. Remaining time was spent watching a neutral video (clips from the BBC TV series Life). Participants then completed activities not relevant to the current study. Four additional saliva samples were taken during this recovery period (immediately as well as 10, 20, and 30 min after the TSST; these post-TSST samples began approximately 15 min after the baseline sample, and were taken while participants were completing questionnaires that were unlikely to affect physiological recovery). Cardiovascu­ lar indexes were monitored every 5 min throughout the recovery period.

Recovery 35m

T S2

T S3

45m

55m

S4

S5

■ T

Visit timeline for youth. Note: S = saliva sample.

1

FAMILY RELATIONSHIPS AND PHYSIOLOGICAL REACTIVITY

Procedure: Parents Parents filled out the original PANAS (Watson, Clark, & Tellegen, 1988) and then engaged in two interactions, the order of which was counterbalanced. For this study, only behavior during the standardized and widely used interaction task (described be­ low) was evaluated; there were no differences in behavior based on which interaction parents engaged in first, Is < 1.05, ps > .30. Couples rated common areas of disagreement, and conflictproducing topics were chosen for them to try to resolve for 15 min (e.g., Kiecolt-Glaser et al., 1996). The discussions were videotaped and later coded (see below) for conflict behaviors.

Assessments of Family Relationships Marital conflict. Because past research suggests that aspects of marital conflict that are described by parents and those that are observed are differentially related to physiological outcomes (Lucas-Thompson, 2009, 2012), we examined both dimensions of marital conflict. Parent-reported marital conflict. Parents reported marital conflict through two questionnaires. The 5-question Conflict sub­ scale from the Braiker-Kelley Partnership Questionnaire (Braiker & Kelley, 1979) was used to measure the frequency and intensity of conflict; the mean of each participant’s responses was calcu­ lated. The resolution subscale from the Kerig Conflicts and Problem-Solving Scales (Kerig, 1996) was used to measure the degree to which partner conflict is successfully resolved. Partici­ pants rated how statements best describe the outcomes of their disagreements. Resolutions are proportionally weighted based on whether they are highly positive (result in increased intimacy, multiplied by 2), unclear or partial (no weighting), or highly negative (result in increased hostility, multiplied by —2). This weighted score was added to the frequency and intensity mean (there was a large correlation between the two self-report mea­ sures, r = .56, p < .001; Cronbach’s alpha: mothers = .79, fathers = .78). An average of both parents’ total conflict scores was utilized (maternal and paternal total conflict scores were significantly correlated, r = .53, p < .001; M = -1 .0 6 , SD = 5.90). Observed conflict behavior. Each parent was rated based on the degree to which specific behaviors were displayed (0 = absent to 2 = very strong display, Lucas-Thompson, 2012). A negativeconflict-tactics score was created by summing ratings of nonverbal and verbal anger, defensiveness, withdrawal, distress, physical aggression, threat, pursuit, insult, and withdrawal (Cronbach’s alpha: mothers = 0.65, fathers = 0.66). Mothers and fathers were coded by different observers. There was adequate reliability prior to consensus coding (ICC = .82). For use in analysis, an average of both parents’ negative conflict behavior was calculated (mater­ nal and paternal conflict behavior were significantly correlated: r = .46, p < .001). This variable displayed significant skew. Therefore, square root transformed values were used in analyses. Relationship quality with parents. Youth completed the Pa­ rental Warmth, Support, and Hostility Scale (Conger et al., 2002). Participants answered 13 questions about their relationship with each parent on a scale from 0 = never to 3 = usually. Relation­ ships with mothers (Cronbach’s alpha = .89) and fathers (Cron­ bach’s alpha = .89) were examined separately because they are

541

often differentially related to child outcomes (e.g., Roubinov & Luecken, 2010). Determination of salivary analytes. Saliva was collected using the Salivette device (Sarstedt, Niimbrecht, Germany). Saliva samples were kept at —20 °C until shipped for analysis. Following Granger, Kivlighan, el-Sheikh, Gordis, and Stroud (2007), saliva samples were assayed for sAA by kinetic reaction assay using a commercially available kit without modification to the manufacturers recommended protocol (Salimetrics, State Col­ lege, PA; sAA was sent to Salimetrics for assaying). The assay employs a chromagenic substrate, 2-chloro-4-nitrophenol, linked to maltotriose. The enzymatic action of sAA on this substrate yields 2-chloro-p-nitrophenol, which can be spectrophotometrically measured at 405 nm using a standard laboratory plate reader. The amount of sAA activity present in the sample is directly proportional to the increase (over a 2-min period) in absorbance at 405 nm. Results are computed in U/mL of sAA. Intraassay variation computed for the mean of 30 replicate tests was less than 7.5%. Interassay variation computed for the mean of average duplicates for 16 separate runs was less than 6%. Before assaying the samples for cortisol (conducted at the University of Trier), thawed samples were centrifuged at 2,000 g for 10 min. Salivary cortisol levels were determined by a solid phase timeresolved fluorescence immunoassay with flouromeric end point detection (DELFIA). The intraassay coefficient of variation was between 4.0% and 6.7%, and the corresponding interassay coeffi­ cients of variation were between 7.1%-9.0%. All samples were assayed in duplicate and averaged. To measure cortisol and sAA production during the stressor, as recommended by Pruessner, Kirschbaum, Meinlschmid, and Hellhammer (2003), both “area under the curve with respect to ground” (AUCg) and “area under the curve with respect to increase” (AUCi) were examined. AUCg represents total sAA/cortisol con­ centration over the course of the visit. AUCi represents change over time, as it is total area under the curve taking into account the first value. This variable represents whether sAA/cortisol levels changed over the course of measurement. Although both were included in line with recommendations (Pruessner, Kirschbaum, Meinlschmid, & Hellhammer, 2003), we expected that AUCi would be more strongly related to family relationship quality because it is a measure of reactivity specifically (vs. total amount of production). AUCg and AUCi were calculated using guidelines in Pruessner et al. (2003). AUCg values were skewed for both sAA and cortisol. AUCi cortisol values were skewed. Analyses on outcomes with skew utilized square root transformed values. Cardiovascular reactivity. During the TSST, BP and HR were monitored every 3 min using a DINAMAP Pro. During the rest of the visit, BP was measured every 5 min. To examine reactivity, average BP and HR were calculated during baseline (5 min and 10 min prior to the TSST) and during the TSST (0-, 3-, 6-, and 9-min post-TSST onset); levels during baseline were sub­ tracted from levels during the TSST (i.e., positive values reflect increases). SBP and HR reactivity displayed significant skew; square root transformed values were used.

Potential Control Variables Age, pubertal status, height, weight, and ethnicity (White vs. non-White) were included as controls because of links in previous

LUCAS-THOMPSON AND GRANGER

542

work with the study variables. Other variables (stressful life events, parental depression, relevant health behaviors [exercise; consumption of caffeine, alcohol, nicotine, food, and medication; hours slept], date of last menstrual period [females only], and other demographic characteristics [gender, income, maternal education]) were included as controls only when they were correlated with family relationships and physiology.

Analytic Plan Primary analyses were conducted separately on sAA (AUCg and AUCi), cortisol (AUCg and AUCi), and cardiovascular (SBP, DBP, and HR) reactivity. Generalized Estimating Equation (GEE) models (regression-based, nonparametric) were used to adjust for clustering of youth within families (e.g., Ballinger, 2004). Because we were investigating multiple outcomes and interactions, p values were adjusted using a false discovery rate methodology (Benjamini & Hochberg, 1995). In addition to height, weight, age, pubertal status, and ethnicity, included as controls were caffeine (sAA AUCg) and chip consumption (DBP); analyses predicting cardiovascular reactivity also controlled for baseline levels. There did not appear to be problematic multicollinearity among predic­ tors. Interactions were tested by creating multiplicative interaction terms (after centering) and controlling for lower-order terms. In­ teractions were tested separately and interpreted following proce­ dures outlined by Aiken and West (1991). Power analyses (using GPOWER; Faul & Erdfelder, 1992) suggested that the GEEs had adequate to excellent power to detect small, medium, and large effect sizes (.69-1.00).

Results Descriptive and Bivariate Patterns of Stress Responses Responses to the TSST. On average, there was a significant quadratic effect of time on sAA over the course of the TSST, F (l,

135) = 4.01, p = .047, with levels increasing from baseline (Tl; M = 43.90, SD = 37.'74) to T2 (M = 55.20, SD = 49.45), decreasing from T2 to T3 (M = 44.58, SD = 40.16), and then staying relatively stable across measurements T3-T5 (T4: M = 45.20, SD = 36.88; T5: M = 42.19, SD = 32.18). Similarly, there was also a significant quadratic effect of time on cortisol produc­ tion over the course of the TSST, F (l, 139) = 28.15, p < .001, with cortisol levels increasing from cortisol measurements 1-3 (Tl: M = 3.68, SD = 3.78; T2: M = 4.65, SD = 4.65; T3: M = 5.95, SD = 6.56) and then decreasing between measurements 3 and 5 (T4: M = 5.14, SD = 6.'34; T5: M = 4.08, SD = 3.81). In addition, there were significant increases from baseline to during the TSST in SBP (baseline: M = 105.63, SD = 10.35; TSST: M = 118.88, SD = 14.54), DBP (baseline: M = 62.90, SD = .64; TSST: M = 73.66, SD = .73), and HR (baseline: M = 77.28, SD = 11.28; TSST: M = 85.45, SD = 13.67), ts > -9 .5 6 , ps < .001.

Bivariate associations. Parent-reported (but not observed) marital conflict was correlated with significantly lower-quality mother-child (but not father-child) relationships (see Table 1). In terms of physiology, there were no bivariate associations between marital conflict and sAA, but parent-reported conflict was associ­ ated with significantly reduced HR reactivity. Higher quality re­ lationships with fathers were correlated with greater sAA AUCi.

Unique “Effects” of Marital Conflict and Parent-Child Relationships Before considering parent-child relationship quality (see Table 2), more frequent, intense, and poorly resolved conflict (as re­ ported by parents) predicted significantly greater increases in sAA as indexed by AUCi. These associations remained once parentchild relationship quality was added. There were significant neg­ ative associations between parent-reported marital conflict and HR reactivity (which were reduced after controlling for parenting).

Table 1 Bivariate Correlations Among Marital Conflict, Parent-Adolescent Relationship Quality, Conflict Appraisals, and SAA 3

2

Marital conflict 1. Parent-reported 2. Observed Parent-adolescent relationship quality 3. Mothers 4. Fathers sAA 5. AUCg 6 . AUCi Cortisol 7. AUCg 8 . AUCi Change from baseline to TSST 9. SBP 10. DBP 11. HR Demographic control variables 12. Age 13. Pubertal status 14. Ethnicity > < .1 0 .

*p
< .0 1 .

4

- . 2 0 * -.1 4 -.0 7 -.0 6 X

.55*** X

5

6

7

.0 2

-.0 4

-.0 1

.06

- .1 0

-.0 8 -.03

.2 1 *

.08

X

.1 0

—,16+ X

.09 .07 .03 - .0 1

X

8

-.0 6 -.0 8

.1 1

.06 -.0 8 -.0 8 .8 8 *** X

9 -.13 -,1 6 +

10

-.0 5 -.03

11

-.26* -.0 4

12

- .1 0

.04

-.0 3 - .1 1

14 -.25** -.26**

.03 .07

.04 .0 1

- .0 2

-.29** -.32***

-.25** -.32***

.0 1

-.1 3 .18+

-.07 -.07

-.19* .15+

.1 2

-.0 9

-.0 6

.1 0

.03 -.07

.19* .2 2 * .2 0 * .08 .26**

.14 .2 2 * .35** X

.24** .1 2

.50*** X

.06

.28** .23** 4 9 *** .48*** X

.09 .1 2

.1 0

.16

.0 1

.0 1 0

.18*

.13

X

* * > < .0 0 1 .

13

.63*** X

.07 .1 2

.16* .04 X

FAMILY RELATIONSHIPS AND PHYSIOLOGICAL REACTIVITY

543

Table 2 Generalized Estimating Equation (GEE) Models Predicting SAA and Cortisol AUCg and AUCi Based on Marital Conflict and Parent-Adolescent Relationship Quality sAA AUCga-b,c Marital conflict Parent-reported Observed Parent-child relationship quality Mothers Fathers

sAA AUCib,c

Cortisol AUCga-b-c

Cortisol AUCib,c

SBPa,b'(

DBPibc

HRab-d -.0 4 d (.03) ,39+ (.21)

.10(31) 3.17 (2.46)

70.23 (15.88)*** .29 (135.61)

-.11 (.13) -.38(1.01)

.06 (.17) -.21 (1.35)

.03 (.03) -.0 5 (.19)

.07 (.12) 1.78+ (1.01)

-5.05(5.18) 1.73 (3.91)

-795.41** (299.32) 802.94** (294.95)

1.02(1.62) 1.09(1.26)

1.43(2.82) -1.29(3.00)

-.0 6 (.32) .18(30)

.20(1.77) 1.05 (1.67)

.33 (.40) .03 (40)

Square-root transformed. Analyses controlled for age, pubertal status, ethnicity, height, and weight; analyses predicting sAA AUCg only also controlled for caffeine consumption, and analyses predicting DBP also controlled for chip consumption. c Coefficients for marital conflict variables were similar before controlling for parent-child relationship quality. d Before controlling for parent-child relationship quality, there was a significant association of parent-reported marital conflict with HR responding: b = -.0 5 , SE = .02, p < 05 *p < .05. **/> < .01. ***p < .001.

These analyses also revealed that, controlling for marital conflict, significantly greater sAA reactivity was displayed by youth who reported less-positive relationships with their mothers but more­ positive relationships with their fathers. All of these main effects were qualified by significant interactions (see below), except links between marital conflict and HR reactivity.

The Moderating Role of Parent-Child Relationship Quality sAA. Mother-child relationship quality moderated associa­ tions between parent-reported marital conflict and both sAA AUCg and AUCi (Figure 2A and B); father-child relationship quality similarly moderated associations of both parent-reported (significantly) and observed negative conflict behavior (margin­ ally, adjusting for false detection rates) with sAA AUCi (Figure 2C). Across interactions, examination of the simple slopes indi­ cated that there was a significantly stronger association between conflict and sAA for those with low-quality relationships with their parents, as compared with those with high-quality relation­ ships. For youth with poor-quality parental relationships, marital conflict predicted heightened sAA production or reactivity. For youth with higher-quality relationships, marital conflict was either not related to sAA or predicted heightened reactivity, but to a lesser extent. Cortisol. There were no interactions between marital conflict and parent-child relationship quality in relation to cortisol pro­ duction. Cardiovascular reactivity. The association between ob­ served negative conflict behavior and SBP change from baseline to the TSST was moderated by father-child (significantly) and mother-child (at trend levels) relationship quality (Figure 2D). In both cases, high levels of negative conflict behavior predicted dampened or reduced SBP reactivity to the TSST, relative to lower levels of negative conflict behavior. However, this association between conflict behavior and dampened reactivity was signifi­ cantly stronger for youth with low-quality relationships with their parents compared with youth with high-quality relationships with their parents.

Discussion This study examined the unique and interactive effects of mar­ ital conflict and parent-child relationship quality in relation to adolescents’ acute stress reactivity to social stress. Results sug­ gested that marital conflict predicted sensitized sAA but dampened HR reactivity. Interestingly, parent-child relationships moderated links between marital conflict and stress reactivity, such that marital conflict predicted heightened sAA or dampened SBP re­ activity to the TSST when parent- child relationships were lacking in warmth and support; when these relationships were of high quality, marital conflict was less strongly related to reactivity. The findings have several noteworthy implications for biosocial mod­ els of the family, as well as theories of the social determinants of biological sensitivity to context. Contemporary theoretical perspectives underscore that the fam­ ily context has the potential to influence individual differences in the set point of biobehavioral responses to stress (Booth et al., 2000). Given past research and theory, we anticipated that char­ acteristics of marital and parent-child relationships would work interactively together to shape individual differences in physiolog­ ical stress responses. Specifically, we hypothesized that highquality parent-child relationships would protect children from potential physiological dysregulation. Results generally supported these hypotheses: typically, marital conflict predicted heightened sAA and dampened SBP reactivity only or more strongly for children with poor-quality relationships with parents. These effects were evident most consistently in relchildren with poor-quality relationships with parentsation to sAA responding. sAA activity is strongly related to other measurements of SNS responding (e.g., Skosnik, Chatterton, Swisher, & Park, 2000); therefore, this pat­ tern of results raises the possibility that sAA responding specifi­ cally or SNS reactivity more broadly may be more strongly related to characteristics of multiple family relationships than responding in other physiological systems. It is typically difficult to develop high-quality relationships with children when parents are drained by the stress of marital conflict (Erel & Burman, 1995), but doing so may buffer children from experiencing potential physiological dysregulation. Thinking of positive relationships as protective is in line with evidence and

LUCAS-THOMPSON AND GRANGER

544

— ♦ -L ow motherchild relationship quality —• — High motherchild relationship quality Parent-reported conflict

Parent-reported conflict

0 Low marital conflict High marital conflict -500

a

-1000

Parent-child relationship quality moderates the link between marital conflict and adolescents' physiological responses to social evaluative threat.

This study examined how marital conflict and parent-child relationship quality moderate individual differences in adolescents' adrenocortical and auto...
7MB Sizes 0 Downloads 3 Views