Psychological and Environmental Correlates of HPA Axis Functioning in Parentally Bereaved Children: Preliminary Findings

Abstract

Parental loss during childhood is a significant public health problem, with the most recent projected estimate of 25 million newly orphaned youth worldwide in 2010 alone (UNAIDS, UNICEF, & USAID, 2004). Symptoms of distress are evident in parentally bereaved children (Goenjian et al., 2009; Kaplow, Saunders, Angold, & Costello, 2010), and a significant minority (approximately 10%; Dowdney, 2000; Melhem, Porta, Shamseddeen, Walker Payne, & Brent, 2011) exhibits clinically disruptive grief reactions. The biological, psychological, and environmental predictors of such reactions are still largely unknown. In the midst of ongoing developmental changes, bereaved children depend heavily on their immediate caretaking environment to help facilitate their mourning (Clark, Pynoos, & Goebel, 1994). Accordingly, efforts to distinguish between features of positive adjustment versus maladjustment following bereavement in youth must address grief and mourning within the broader context of both individual (e.g., biological mechanisms, coping strategies) and socioenvironmental (e.g., parental psychopathology) factors that may diminish, promote, or serve as markers of these outcomes (Kaplow, Layne, Pynoos, Cohen, & Lieberman, 2012). The identification of biological and psychosocial mechanisms linking childhood bereavement to psychiatric symptoms will help to inform the development of targeted interventions for at-risk bereaved youth.

The hypothalamic–pituitary–adrenal (HPA) axis and its end product, cortisol, play a key role in mediating the negative impact of stress on the onset, course, and pathophysiology of psychiatric disorders (Adam & Kumari, 2009; Pfeffer, Altemus, Heo, & Jiang, 2007) and could serve as a mediator of or risk marker for maladaptive responses to early parental loss. Short-term elevations in cortisol are critical for survival under stressful circumstances, but prolonged exposure to high cortisol levels appears to be associated with serious health risks (Bevans, Cerbone, & Overstreet, 2008). Hypercortisolemia or impaired feedback control of the HPA axis, for example, are well-established biomarkers of depression, in both adults (e.g., Gold, Goodwin, & Chrousos, 1988; Young, Haskett, Murphy-Weinberg, Watson, & Akil, 1991) and children/adolescents (e.g., Kaufman et al., 1997; Rao et al., 1996). A growing body of evidence, however, suggests that blunted reactivity in this system also may play a role in the pathophysiology of stress-related disorders (for review see Gunnar & Vasquez, 2001).

“Hypocortisolism” can be characterized by low cortisol secretion, flat cortisol production patterns during the day, or blunted cortisol responses to stressors (Heim, Ehlert, & Hellhammer, 2000); all of these have been associated with exposure to early life stress. Early life adversity may provoke frequent elevations in cortisol that lead to compensatory downregulation and reduced activity of the HPA axis in later life (Gunnar & Vasquez, 2001). Others (e.g., Hellhammer & Wade, 1993; McEwen 1998) have similarly argued that hypocortisolism can reflect allostatic load, defined as the cumulative burden and physiological consequences of chronic exposure to heightened neuroendocrine response that results from chronic stress. For example, a history of repeated trauma exposure in youth (higher levels of both previous and recent trauma, including the death of a loved one) is associated with blunted morning cortisol levels (Bevans et al., 2008). A poorly reactive stress system, perhaps created by early adversity exposure, may then contribute to psychiatric vulnerability. Consistent with this view, lower morning cortisol levels and a blunted cortisol response have been associated with poor parental responsiveness, a neglectful environment, and insecure attachment patterns (Carlson & Earls, 1997; Spangler & Grossman, 1993) during infancy and early childhood, all of which may contribute to psychopathology risk.

Efforts to link HPA axis dysfunction directly to psychiatric symptoms in youth have produced variable results. High cortisol levels were positively associated with PTSD symptoms in trauma-exposed children, but only when the trauma was recent (past year); with prior trauma this relationship disappeared (nonsignificant) and its directionality was reversed (Weems & Carrion, 2007). Low morning cortisol levels were also associated with increased PTSD symptoms in an Armenian earthquake study (Goenjian et al., 1996). The very few studies of parentally bereaved children that have examined links between cortisol and psychiatric symptoms show a similar hypocortisolism response. A greater number of negative life events following the loss of a parent predicted lower cortisol levels (measured before, during, and after a conflict discussion task with the surviving parent) 7 years later (Hagan, Luecken, Sandler, & Tein, 2010). In a retrospective study of college students, early loss experience was associated with decreased salivary cortisol response to awakening (Meinlschmidt & Heim, 2005).

Inconsistencies in the extant literature are not surprising given that published studies have examined very different types of stressors (e.g., natural disasters, abuse, parental loss)—in samples that have differed in age, gender, time since trauma and duration of trauma—and have used differing measures of HPA axis function (e.g., mean cortisol levels, cortisol awakening response, diurnal decline, challenge reactivity; Adam & Kumari, 2009; Gunnar & Vasquez, 2001). The ability to obtain cortisol data noninvasively has made studies in children much more feasible, but there has been a tendency to treat disparate HPA axis measures equivalently and to regard “hyper” or “hypo” activity simplistically. Extensive use of salivary cortisol measures in epidemiological work has increasingly focused on the cortisol awakening response (CAR) and diurnal decline, which can be efficiently obtained by sampling upon awakening, 30 minutes later, and in the evening (Adam & Kumari, 2009). The normal circadian rhythm of cortisol produces rising levels in the early morning hours, a further abrupt rise that commonly leads to the highest levels of the day 30 to 45 minutes after awakening, and falling levels through the rest of the day to an evening nadir. The CAR, calculated as the magnitude of rise from awakening to 30 minutes thereafter, appears to be influenced by different regulatory processes than the remaining diurnal profile (Rosmalen et al., 2005), and it likely provides different information about the stress system than do other measures of HPA reactivity. We chose to focus our analyses on the CAR because it has been the component of cortisol’s circadian rhythm most extensively examined in psychosocial stress studies, and because it has been standardized in a way that facilitates comparison across studies.

Despite variation in methods and results, a growing body of evidence supports the idea that reduced stress reactivity may be associated with more maladaptive outcomes. Hypocortisolism has been linked in adults with avoidant coping strategies (Tull, Sheu, Butler, & Cornelious, 2005) and a history of parental psychopathology (Luecken & Appelhans, 2006). In trauma-exposed children (including those exposed to parental loss), avoidant coping and/or emotional suppression (Kaplow, Dodge, Amaya-Jackson, & Saxe, 2005; Kaplow, Gipson, Horwitz, Burch, & King, in press; Shapiro, Kaplow, Dodge, & Amaya-Jackson, 2012) as well as poor functioning of the surviving caregiver (Lin, Sandler, Ayers, Wolchik, & Luecken, 2004) predict psychological distress. Each likely factors into the relation between grief and HPA axis activity, but these complex relations have yet to be examined in parentally bereaved youth.

This study examined bereaved children’s HPA-axis functioning, as measured by the CAR, in relation to psychological distress, coping, and the surviving parent’s grief response. We hypothesized that greater child psychiatric symptoms, child avoidant coping, and parental psychopathology would each be associated with a dampening of child CAR. To our knowledge, this is the first study to focus specifically on the CAR as a measure of HPA axis functioning in relation to both psychological and environmental factors in parentally bereaved children following recent parental death.

Method

Results

Descriptive data for cortisol and other variables are presented in Table 1. Pearson’s correlations and t tests showed that race, age, life stress, the cause of death, and length of time since loss were not associated with our primary variables of interest, so we did not control for them in further analyses. Although girls exhibited higher levels of PTSD symptoms than boys, t(33) = −2.35, p = .025, no gender differences were noted among any of the other variables. Bivariate analyses showed very little association between children’s Day 2 or Day 3 CAR and symptom domains of interest.

Results (see Table 2) showed significant relations between a dampening of child CAR (reduction in slope between a.m. wake-up and 30 minutes later) and more symptoms of child posttraumatic stress, F(1, 22.6) = 7.10, p < .05; anxiety, F(1, 25.19) = 6.26, p < .05; maladaptive grief, F(1, 22.67) = 7.07, p < .05; and depression, F(1, 25.92) = 4.93, p < .05. A dampening of CAR was also associated with children’s greater use of avoidant coping, F(1, 22.47) = 10.06, p < .01, as well as more parent maladaptive grief symptoms, F(1, 26.13) = 9.39, p < .01.

Post hoc analyses searched for links between psychological measures and absolute levels of cortisol at awakening, 30 minutes later, and in the evening. Even using loose criteria (by not correcting for multiple comparisons), none of these correlations were significant, except for a few with the +30 minute sample, which were redundant with the CAR associations because this also captures the response to awakening.

Discussion

Findings support links between avoidant coping, child psychological distress, parental maladaptive grief, and a dampened CAR in parentally bereaved children. These findings are consistent with other studies that have viewed hypocortisolism as a marker of maladaptation. Hypocortisolism can reflect allostatic load (Hellhammer & Wade, 1993; McEwen 1998) and appears to develop in some youth following prolonged periods of elevated cortisol in response to trauma or extended periods of general life stress. Given that the death of a parent is not an isolated incident (e.g., parental loss often occurs in the context of other significant life stressors; Kaplow et al., 2010), may have been anticipated for months or years (i.e., 34% of our sample experienced an anticipated death), and is usually followed by numerous secondary adversities (Hagan, Luecken, Sandler, & Tein, 2010), the current study’s findings may indeed support the idea that accumulating allostatic load can lead to blunted HPA axis reactivity, at least as reflected in the CAR.

Follow-up, longitudinal work is needed to test the hypothesis that accumulated burdens of early trauma/stress can alter set points in stress response systems that contribute to failures in normal system activation and subsequent development of psychiatric symptoms. There are numerous potential pathways through which this could occur including compromised parent– child communication, which is often identified in stressed and/or depressed mothers (Shapiro et al., in press; Yehuda, Halligan, & Bierer, 2001), predicts both parents’ (Granger et al., 1998) and children’s cortisol levels (Smeekens, Riksen-Walraven, & van Bakel, 2007), and may be more likely to occur when individuals employ avoidant coping strategies. Our own prior work suggests that in trauma-exposed children (including those exposed to parental loss), avoidant coping predicts psychological distress (Kaplow et al., 2005, in press; Shapiro et al., 2012). Here we show that avoidant coping also predicts a blunted CAR. Others have demonstrated that poor functioning of the surviving caregiver (Lin et al., 2004) predicts psychological distress in the child, and here we show that this also predicts blunted child CAR. The causal path may lead from parenting factors that contribute to poor coping with loss within some families (e.g., avoidant coping responses, poor communication) and together contribute to accumulating allostatic load and reduced HPA axis reactivity. Our data, however, cannot tell us if the altered CAR is a biomarker of pre existing vulnerability, or reflects a mediating pathway from early life stress exposure to psychiatric symptom expression. Thus, larger, longitudinal studies are needed.

Another explanation for blunted CAR is suggested by our finding that the associations detected were consistent across measures for Day 1 CAR, but largely disappeared for Day 2 and Day 3 CAR. Much of the literature has framed HPA axis biomarkers as predictors of clinical outcomes, but our results raise the possibility that blunted CAR seen in the more troubled children was a result of stressful and emotionally challenging events on the interview day. Interviews asked children to reveal thoughts and feelings about the deceased parent. Although not an explicit part of the current study, the assessment concluded with a parent–child interaction task during which they were asked to talk together about positive memories of the deceased as well as characteristics that the child may have had in common with the deceased parent. How could these events or their subsequent processing within the family (or lack thereof) impact next-day CAR and not have a similar impact on Day 2 and Day 3 CAR?

The CAR may primarily reflect adrenal responsivity (Schmidt-Reinwald et al., 1999) rather than more central aspects of HPA regulatory control, and may be shaped by both neural processes associated with the awakening cycle (Clow, Hucklebridge, Stalder, Evans, & Thorn, 2010) and psychological processes occurring during sleep (Born, Hansen, Marshall, Molle, & Fehm, 1999). Evidence suggests that CAR may be most strongly shaped by state factors (Armario,Valles, Dal-Zotto, Marquez, & Belda, 2004; Stalder, Evans, Hucklebridge, & Clow, 2010). A recent review of CAR neurobiology (Clow et al., 2010) highlights its inconsistent relationship with psychosocial and health variables, emphasizes the role of extrahypothalamic factors in shaping its cortisol rise, and suggests its most critical role may be in the transition from sleep to wakefulness, with significant regulation by the suprachiasmatic nucleus through neural modulation of adrenal sensitivity to adrenocorticotropic hormone. This response appears to be carefully orchestrated by interacting systems managing the transition from sleep. Consistent with this view, chaotic or unpredictable sleep/wake cycles have been associated with blunted daytime cortisol pattern for some populations of children (Gunnar & Vasquez, 2001).

It is possible that the content of the interview itself, especially the parent–child interaction in which discussions regarding memories of the deceased person were the focus, induced distress primarily for those children exhibiting avoidant coping patterns and experiencing parental maladaptive grief, which may have contributed to disrupted sleep and subsequent blunting of the next-day CAR. Behavioral coding of the dyadic parent–child interviews from this sample indeed suggests that children’s observed discomfort during the interaction predicts next-day CAR blunting (Shapiro, 2012), and other studies have found that parent–child discussions may produce declining cortisol patterns (Adam, Klimes-Dougan, & Gunnar, 2007). The fact that the only significant correlate of Day 2 or 3 CAR is parental maladaptive grief is consistent with the idea that the interview itself may have served as a standardized trauma or loss reminder (Layne et al., 2006), which created more temporary disruptions in those families already struggling with maladaptive grief reactions, perhaps contributing to sleep disruption and CAR blunting.

Existing evidence strongly supports the idea that prior-day emotional experiences significantly shape next-day CAR (Stalder et al., 2010). Further work is needed to determine whether this occurs secondarily through effects on sleep, or through an interaction between acute emotional experiences and neuroendocrine control system set points shaped by prior stress exposures. The latter is more consistent with the accumulating allostatic load hypothesis proposed earlier, wheras sleep mediation suggests a different mechanism. Both factors may be in play, and each may be more strongly reflected in different measures of HPA axis function. Further work exploring both of these possibilities must take into account the complexity of the biology involved in HPA axis functioning.

Acknowledgments

References