Abstract
This study explored the relationship between individual and family-level risk in predicting longer-term parental distress following their child’s unexpected diagnosis of serious illness. A mediation model was tested, whereby parents’ pre-existing psychosocial risk predicts longer-term posttraumatic stress symptoms, indirectly through parents’ acute stress response. One hundred and thirty-two parents of 104 children participated. Acute stress response partially mediated the relationship between psychosocial risk and posttraumatic stress symptoms, with a moderate indirect effect (r2 = .20, PM = .56, p < .001). Findings demonstrated that cumulative psychosocial risk factors predispose parents to acute stress and longer-term posttraumatic stress symptoms, highlighting the need for psychosocial screening in this population.
Introduction
A diagnosis of life-threatening pediatric illness can have an intense and disruptive impact on children and their families. As a result, many parents experience acute psychological distress and longer-term posttraumatic stress symptoms (PTSS). PTSS may arise as a protracted response to a traumatic event, occurring at least 4 weeks after the event, and persisting for months to years later (American Psychological Association, 2015). There is robust evidence demonstrating elevated rates of clinically significant PTSS in parents in response to range of illnesses, including cancer diagnoses (McCarthy et al., 2012; Pöder et al., 2008), injuries from road traffic accidents (Allenou et al., 2010; De Vries et al., 1999), admission to the pediatric intensive care unit (PICU; Bronner et al., 2010), and organ transplantation (Rossi et al., 2011). Collectively, these studies highlight the substantial proportion of parents experiencing ongoing PTSS, across a range of life-threatening illnesses and medical settings (Best et al., 2001; Hall, 2007; Kazak et al., 2004; McCarthy et al., 2012; Winston et al., 2003). Common to these medical events are the parents’ perceptions of their child experiencing significant pain (Kassam-Adams et al., 2009), and the degree of concern that they might die (Balluffi et al., 2004; Kassam-Adams et al., 2009).
PTSS is characterized by a presentation of event-related re-experiencing and persistent avoidance, and increased arousal. The symptom severity of PTSS falls below the threshold for formal posttraumatic stress disorder (PTSD) diagnosis (Best et al., 2001); however, sufferers exhibit a similar profile of symptomatology. The experience of clinically significant PTSS has implications for both the parent and the ill child, with parents’ level of PTSS correlating with their own quality of life (Kazak et al., 1997), their physical and psychological wellbeing (Mowery, 2011), as well as their child’s level of distress (Morris et al., 2012), psychopathology (Leen-Feldner et al., 2011), and emotional development (Marsac et al., 2011; Seng et al., 2005). The effects of childhood illness on parent mental health are of significant clinical importance; yet the complexity of individual and family-level risk factors is yet to be well understood.
The model of Pediatric Medical Traumatic Stress (PMTS) presented by Kazak (2006) uses a social ecological framework to describe the trajectory and recovery of parents’ experience of acute psychological distress, and organizes risk factors in a biopsychosocial framework. In this model, psychosocial factors are presented as being important predisposing factors for parents’ acute stress responses, with the severity of this early distress response proposed to predict the course of longer-term PTSS. Psychosocial (individual and family-level) risk factors include the parents’ capacity to cope with stress, perceived social support, family and life stressors, financial strains, family structure, and family beliefs (Kazak et al., 2003). These factors have been found to correlate with parents’ acute stress responses (McCarthy et al., 2012) as well as longer-term PTSS (Brewin et al., 2000; Hall et al., 2006; Horsch et al., 2012; Kazak et al., 1997) in a range of pediatric settings. Importantly, although this model was originally developed and informed by research within the oncology setting, it suggests that these parental traumatic stress responses occur across different illness groups and proposed that this model can be extrapolated to other illnesses.
While the linear relationships between these factors have been examined, it is still unclear why only a proportion of at risk parents experience ongoing PTSS. It is possible that psychosocial risk may predispose parents to ongoing PTSS, with this relationship mediated by the severity of the stress responses experienced in the acute period. This relationship is consistent with the PMTS model (Kazak, 2006; Price et al., 2016), however is yet to be reported. This study will therefore further explore and characterize the nature of the relationship between psychosocial risk factors and the development of longer-term PTSS, beyond the correlations found within the literature. The study will also provide further empirical data to support the relationship theoretically presented in the PMTS model, in which the development of PTSS is not illness specific, and occurs across illness groups.
This study expands on current findings in the parent trauma literature. It was specifically aimed to explore the impact of child illness on the acute stress responses in parents, and the development of later PTSS. Consistent with previous literature (Allenou et al., 2010; McCarthy et al., 2012), it was hypothesized that there would be non-statistical differences for parents across child illness groups. Using a mediation model, the study also aimed to examine the relationship between psychosocial risk, acute stress, and longer-term PTSS. In accordance with the PMTS model (Kazak, 2006; Price et al., 2016), it was hypothesized that parents with a greater number of pre-diagnosis psychosocial risk factors would experience a higher level of acute stress response (ASR) within the 4 weeks following their child’s diagnosis of life-threatening illness. Furthermore, it was expected that higher levels of ASR would predict more significant PTSS 4 months later, thereby mediating the relationship between psychosocial risk and PTSS.
Methods
Study design and participants
Results are from time point one (T1: within the first 4 weeks since admission/diagnosis) and time point two (T2: 4 months following admission/diagnosis) of a larger prospective longitudinal study (The Take a Breath study at The Royal Children’s Hospital, Melbourne, Australia).
Participating parents (99 women, 33 men, Mage = 36.2 years, age range: 19–77 years) of children (42 girls, 62 boys, Mage = 2.7 years, age range: 0–15 years) were recruited from first admissions to The Royal Children’s Hospital Oncology, Cardiology, and PICU departments between November 2010 and August 2012. Parent and children demographic and illness group details are summarized in Table 1. With regard to specific diagnoses, within the Oncology group, 37 percent of patients had a type of leukemia, 45 percent had solid cancers, and 18 percent had brain cancers. Within the Cardiology group, 35 percent underwent biventricular repair (e.g. Tetralogy of Fallot and transposition of the great arteries), 13 percent underwent single ventricle repair (hyperplastic left heart syndrome), and 52 percent underwent other surgeries. Within the PICU group, 34 percent had cardiac disease, 29 percent had a severe respiratory illness, 17 percent sustained a trauma, 11 percent had an infectious disease, and 9 percent were admitted for other serious illnesses. Figure 1 shows sample recruitment details for the present study. Local privacy legislation does not allow for comparison of participants and non-participants; therefore, further examination of the representativeness of the sample was not possible. Participants received standard medical care, including access to a social worker and other allied health staff members.
Demographic characteristics of parents and children across diagnosis groups.
PICU: pediatric intensive care unit.
N = 132.
Participant numbers at each stage of the study.
Inclusion criteria were parents having an active and current parenting role with the hospitalized child, who is under 18 years of age, and diagnosed with or admitted with their first presentation of life-threatening illness within the previous 4 weeks. Parents were excluded if they were under the age of 18 years, had limited spoken English and/or literacy, or if they reported having experienced another significant trauma in the 2 months prior to or concurrent with the child’s illness.
Procedures
This research was conducted at The Royal Children’s Hospital, Melbourne, Australia, and was approved by The Royal Children’s Hospital Human Research Ethics Committee (HREC 30044). Research team members monitored daily admissions lists of the departments, and liaised with clinical staff to identify eligible families. Written informed consent was obtained and T1 questionnaires were provided to parents for return within 4 weeks of their child’s admission or surgery. Parents were contacted again 4 months after their child’s admission for collection of T2 questionnaires.
Measures
Psychosocial risk
The Psychosocial Assessment Tool (PAT; Pai et al., 2008) is a measure of psychosocial risk factors designed for families with a child newly diagnosed with cancer. It contains seven subscales: family structure and resources, family social support, sibling problems, family problems, parent stress reactions, and family beliefs. Total PAT scores are computed as the sum of subscale scores (range: 0–7), where higher scores indicate greater psychosocial risk. A total PAT score less than 1 represents the lowest risk (universal) category, a score of 1–2 is classified as the elevated (targeted) risk category, and a score greater that 2 represents the highest risk (clinical) category. Internal consistency for the current study was α = .77. Although designed for the oncology setting, the PAT has been successfully used in other illness groups such as kidney transplant (Pai et al., 2012) and cardiac disease (Hearps et al., 2013). The PAT was completed at T1.
Acute stress response
The Acute Stress Disorder Scale (ASDS; Bryant et al., 2000) is a 19-item self-report measure which assesses acute stress in individuals in the acute period (up to 4 weeks) following a traumatic event and identifies individuals at risk for developing PTSD. The ASDS utilizes a 5-point Likert scale and items center around four clusters of symptoms: dissociation, experiencing, avoidance, and arousal. A higher score indicates greater magnitude of ASR symptoms (range: 19–95). The internal consistency of the ASDS total score for the current study was α = .89. The ASDS was completed at T1.
PTSS
The Posttraumatic Stress Disorder Checklist—Specific Incident Version (PCL-S; Weathers et al., 1993) is a 17-item, self-report measure that corresponds to three symptom cluster groups: re-experiencing, arousal, and avoidance. The PCL-S utilizes a 5-point Likert scale. Higher scores indicate greater PTSS burden (range: 17–85). Compared with the Clinician Administered Scale, the PCL-S is both sensitive (se = .94) in detecting PTSD and specific (sp = .89) in discriminating PTSD from other Axis I disorders (Blanchard et al., 1996). The PCL-S total score had a high internal consistency in the current study (α = .92). For this study, parents were asked to complete the questionnaire in relation to their child’s illness. The PCL-S was completed at T2.
Statistical analysis
Data analyses were conducted using the Statistical Package for Social Sciences for Windows (SPSS) Version 21. Preliminary examination of the dataset revealed no obvious outliers, and the overall level of missing items for all data collected was 8.2 percent. Little’s Missing Completely At Random test revealed a non-significant result (χ2(11) = 10.51, p = .486) indicating that data were missing completely at random.
No systematic biases were present for missing data (p > .05), except for parent sex, whereby male parents were more likely to be missing T2 questionnaires (χ2(N = 222, df = 1) = 6.86, p = .009). Cases missing data on entire questionnaires were removed (n = 90). Data missing at the item level were replaced: with 0 for the PAT (i.e. not endorsed) and with a mean weighted score of the relevant subscale for the ASDS and PCL-S. Imputation was minimal for both the ASDS and PCL-S (0.2% and 0.5%, respectively). Natural log transformations were applied to skewed datasets PAT and PCL-S (denoted as PATln and PCL-Sln). Family-clustered correlations were carried out for all combinations of PATln, PCL-Sln, and ASDS total and subscales. Between-illness group comparisons were conducted for ASDS, PATln, and PCL-Sln and were adjusted for within-family relatedness of parents using mixed model analyses, with standardized regression coefficients (β) presented. To test the hypothesized mediation model, mediation analysis as described by Baron and Kenny (1986) was used, employing within-family clustering of parents (Krull and MacKinnon, 2001), and using Sobel’s test (Sobel, 1982) with nonparametric bootstrapping based on 1000 samples (Barrett, 2007).
Results
Demographics
Analyses focus on participants’ ASDS score (M = 43.06, SD = 13.48, range: 19–82), PAT score (M = 0.74, SD = 0.57, range: 0–2.83), and PCL-S score (M = 26.96, SD = 9.88, range: 17–73). Applying recommended screening cut-off score for primary care settings (Andrykowski et al., 1998; Bryant et al., 2000; Dobie et al., 2002), 84.8 percent of the sample demonstrated clinically significant levels of PTSS at time 1, and 27.3 percent at time 2. Mixed models revealed non-significant differences across the illness groups for ASDS score (Wald χ2(2) = 0.24, p = .885), PATln score (Wald χ2(2) = 1.94, p = .379), and PCL-Sln score (Wald χ2(2) = 2.06, p = .356). As our study aimed to include parents from the same family, correlations between same-family mothers and fathers (n = 56) were examined. PAT total scores were positively correlated (r = .48, p = .009), as were ASDS scores (r = .66, p < .001). The outcome variables, PCL-S scores, were not significantly correlated (r = .33, p = .091).
Correlation analyses of study variables across the sample, which can be seen in Table 2, revealed statistically significant relationships (r = .35 to r = .85, p < .001). A higher total score on either PAT or PCL-S was associated with a higher subscale and total score on the ASDS, and a higher total score on the PAT was also associated with a higher total score on the PCL-S.
Correlation among study variables ASDS, PAT, and PCL-S.
ASDS: Acute Stress Disorder Scale; PAT: Psychosocial Assessment Tool (log transformed); PCL-S: Posttraumatic Stress Disorder Checklist—Specific Incident Version (log transformed).
N = 132. All correlations clustered by parent dyad, and p < .001.
Mediation conditions
PATln scores explained a significant proportion of variance in PCL-Sln (R2 = .36, Wald χ2(1) = 70.04, p < .001) and ASDS (R2 = .34, Wald χ2(1) = 57.31, p < .001), thereby meeting the first (path a) and second (path c) conditions of a mediating relationship. Furthermore, ASDS explained a significant proportion of variance in PCL-Sln scores (R2 = .28, Wald χ2(1) = 46.46, p < .001), thereby meeting the third mediating condition (path b).
Indirect effect
For the present dataset, the direct effect of PATln score on PCL-Sln was significant (R2 = .46, Wald χ2(1) = 98.63, p < .001), when the mediator (ASDS score) was included in the model. This effect was weaker than the direct relationship previously established between PATln score and PCL-Sln (β = 0.34, 95% confidence interval (CI) (0.12, 0.55) vs β = 0.60, 95% CI (0.44, 0.76)). This is evidence of partial mediation (PM = 0.56, p < .001), with a moderate indirect effect (r2 = .22). Figure 2 illustrates the mediation model.
Partial mediation model, with single predictor (PAT), mediator (ASDS), and outcome variable (PCL-S). Mediation occurs as the direct effect c’ < c.
Sobel’s test
Sobel’s test was used to confirm whether the mediator (ASDS) carries the influence of PATln score to the outcome variable PCL-Sln. Using this test revealed a statistically significant indirect effect (p < .001). Results supported previous tests, indicating that partial mediation is significant at the 95 percent bias corrected and accelerated CI (z = 4.29; 95% CI = (2.81, 5.74)).
Discussion
The aim of the present study was to expand on current findings in the parent trauma literature, and on our current understanding of parents’ psychological experience in the acute period of their child’s diagnosis of life-threatening illness. Specifically, it aimed to explore whether the type of illness was associated with PTSS levels in parents and to examine the relationship between psychosocial risk, the ASR, and longer-term PTSS.
No significant differences were found in ASR and PTSS in parents across the illness groups, suggesting that illness variables and objective medical factors do not have a significant impact on parent ASR and PTSS. This is consistent with previous literature. Findings also showed that 84.8 percent of parents demonstrated clinically elevated ASR at 4 weeks following their child’s diagnosis, with PTSS evident in 27.3 percent of this group at 4 months post-diagnosis, supporting that medical events are not uniformly traumatogenic across all families. Furthermore, overall the sample fell within the lowest risk category with regard to the Total PAT Score. This finding is consistent with previous research using the PAT, which has found that the majority of the general population fall within this universal risk category (McCarthy et al., 2016; Pai et al., 2008). To explore factors contributing to parent risk of longer-term symptoms, a mediation model of parental distress was examined, which explored the relationships between ASR, psychosocial risk, and PTSS 4 months later. As expected, greater psychosocial risk predicted higher PTSS 4 months later, indirectly through parents’ level of ASR.
Where previous studies have investigated the relationship between psychosocial risk and ASR (McCarthy et al., 2012), psychosocial risk and PTSS (Brewin et al., 2000; Hall et al., 2006; Horsch et al., 2012; Kazak et al., 1997; Pai et al., 2008), and ASR and PTSS (Allenou et al., 2010; McCarthy et al., 2012), this is one of the first studies to examine relationships among these domains within a mediation model. The present findings are consistent with the theoretical perspective presented in the PMTS model (Kazak, 2006; Price et al., 2016). The indirect relationship between psychosocial risk and PTSS through ASR points toward the importance of individual responses embedded within a family unit.
Theoretical implications
The significant role that psychosocial factors play in the ongoing experience of PTSS bolsters existing evidence for a social ecological perspective to trauma. It also supports the PMTS model (Kazak, 2006) as a framework for describing the trajectory of PTSS. Consistent with previous reports, psychosocial risk predicted parents’ level of ASR (McCarthy et al., 2009, 2012). While it may not be the sole cause for ASR and PTSS, the findings highlight psychosocial risk as a family-level modifiable risk factor. Notably, PTSS between parents of the same child was not correlated 4 months after the medical diagnosis, suggesting that other risk factors are also involved, and in keeping with the differing experiences and access to resources of mothers and fathers. This is consistent with previous literature (Kazak et al., 2005) and illustrates the need for further exploration of variables beyond the family context.
Similar to previous studies, the present results suggest that parents’ acute stress response is not explained by the specific medical aspects of their child’s illness. This was despite pronounced differences in age at diagnoses, experience of treatment, follow-up, and remission period that exist across the cardiac, cancer, and PICU groups. Common to all parents was an imposed, and unexpected, threat to their child’s life. This commonality is what may precipitate parents’ acute and ongoing experience of PTSS. The findings suggest that trauma models that include families across multiple life-threatening pediatric illnesses are applicable and relevant. The PMTS model can be applied across medical contexts, to better understand parent response to child illness, and as a useful framework for describing acute pediatric illness.
Clinical implications
The current study findings suggest that medical variables or the child’s illness was not associated with longer-term PTSS in parents. Hence, applying a traumatic stress framework to life-threatening conditions beyond the cancer setting may initiate more widespread trauma-informed medical care for parents and families. Improved identification and treatment of PTSS would improve parent mental health and wellbeing, as well as the psychological and adjustment outcomes of the ill child during their recovery. Given the relationship between family-level risk factors and the development of PTSS in parents, the findings indicate that relevant early screening of parents by healthcare professionals may be warranted. Screening will assist not only in the identification of those experiencing early clinically significant distress levels, but also those with greater psychosocial risk factors, potentially making them more susceptible to the development of PTSS throughout their child’s treatment and recovery period.
Furthermore, intervention research to date has largely focused on families of cancer patients, and supportive data are limited (Kazak et al., 1999; Lewis et al., 1990; Sahler et al., 2002). The finding that the levels of the ASR were similar across the three patient groups suggests that a generalized approach to psychological diagnosis and intervention for parents across acute pediatric settings may be appropriate. The findings also suggest that these interventions could target modifiable psychosocial risk factors to prevent the development of longer-term PTSS.
Limitations and future directions
A large proportion of eligible parents were not recruited in this study, and there was a high rate of participant attrition. Therefore, the sample may not adequately represent the target population. Participant versus non-participant comparison was not possible due to privacy limitations; hence, representativeness of the sample is unknown. One of the strengths of this study was the longitudinal prospective design, and while the high data attrition limits its statistical rigor, it is also reflective of the particular vulnerability of the population of interest. In addition, there is a wide age range of children involved in this study, and the impact of child age was not examined. Despite this limitation, it has consistently been found that child age is not associated with parent distress reactions and the development of PTSS (Woolf et al., 2015).
To maximize parent involvement, both parents from families were included where possible. The results revealed that while psychosocial risk and ASR scores were correlated, parents’ PTSS scores were not. Few studies have examined mechanisms of inter-familial stress response; however, given the strong correlation between parents’ ASR scores, this area warrants further investigation. The findings suggest that over time, the medical event is experienced differently across family members. Examining risk and protective factors for ASR and PTSS in the child with the illness, their siblings, and their parents would provide valuable insight into the family response to medical trauma, as well as support and inform the development of family-orientated therapies.
Addressing the issue of family psychological distress in the acute hospital setting has widespread implications for medically unwell children and their families, and for hospital policies regarding psychosocial screening. These findings support the theoretical PMTS model proposed by Kazak and colleagues in this setting. Furthermore, they indicate that a significant proportion of parents will experience clinical levels of acute stress and later post trauma symptoms, which need to be recognized and supported by healthcare models. Future research is needed to characterize acute psychological responses of families impacted by serious childhood illnesses. This may include the interaction of individual and family risk factors within the PTMS framework, as well as maintaining factors involved in the acute period. This may aid in the development of effective, empirically based clinical treatments for this cohort.
Footnotes
Acknowledgements
The authors acknowledge the generosity of the participating families along with the contributions to the research made by Simone Hearps, Anica Dimovski, Nathan Dowling, and Amy Coe. The larger project was a collaboration with the Parenting Research Centre.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and international committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research project reported in this published work was conducted by the authors as part of their employment with the Murdoch Children’s Research Institute, The Royal Children’s Hospital Melbourne, Australia. The authors acknowledge the generous funding support for this project from the Pratt Foundation and the Victorian Government’s Operational Infrastructure Support Program, and the Royal Children’s Hospital. V.A.A. was funded by a NH&MRC Senior Practitioner Fellowship. None of these funders had any involvement in the planning, design, or execution of this research, and neither played a role in drafting the manuscript. There are no financial relationships or issues to disclose. All authors had full access to the data and can take full responsibility for the integrity of the data and the accuracy of the data analysis.
