Prevalence and associated factors of post-traumatic stress disorder in parents whose infants have single ventricle heart disease

Introduction: Post-traumatic stress disorder occurs in parents of infants with CHD, contributing to psychological distress with detrimental effects on family functioning and well-being. We sought to determine the prevalence and factors associated with post-traumatic stress disorder symptoms in parents whose infants underwent staged palliation for single ventricle heart disease. Materials and methods: A large longitudinal multi-centre cohort study evaluated 215 mothers and fathers for symptoms of post-traumatic stress disorder at three timepoints, including post-Norwood, post-Stage II, and a final study timepoint when the child reached approximately 16 months of age, using the self-report questionnaire Impact of Event Scale – Revised. Results: The prevalence of probable post-traumatic stress disorder post-Norwood surgery was 50% of mothers and 39% of fathers, decreasing to 27% of mothers and 24% of fathers by final follow-up. Intrusive symptoms such as flashbacks and nightmares and hyper-arousal symptoms such as poor concentration, irritability, and sudden physical symptoms of racing heart and difficulty breathing were particularly elevated in parents. Higher levels of anxiety, reduced coping, and decreased satisfaction with parenting were significantly associated with symptoms of post-traumatic stress disorder in

difficulty concentrating, an altered sense of time and place, detachment from self and others, flashbacks, nightmares, difficulty sleeping, and sudden intense fear accompanied by distressing physical symptoms (e.g., racing heart, shortness of breath). 6 formal diagnosis involves a lengthy structured interview with a trained healthcare professional, but shorter self-report screening tools such as the Impact of Event Scale -Revised are frequently used to identify individuals with probable post-traumatic stress disorder. 7n the Family Adaptation study, an ancillary study to the Pediatric Heart Network Single Ventricle Reconstruction Trial, 8 it was reported that parents' perception of their well-being was worse if parents experience increased symptoms of post-traumatic stress disorder. 2Therefore, we sought to further evaluate the prevalence of symptoms of post-traumatic stress disorder in both mothers and fathers of infants with critical CHD who underwent staged surgical palliation.We also investigated potential associations of demographic, clinical, stress, and coping variables with symptoms of post-traumatic stress disorder.

Design of the study
The Family Adaptation study was ancillary to the multi-institutional Pediatric Heart Network Single Ventricle Reconstruction Trial, both previously described. 2,8The Family Adaptation study investigated parents' perceptions of family function, quality of life, and well-being after their infant underwent the Norwood procedure. 2The study theoretical framework was based on the Resiliency Model of Family Adaptation. 9Parents completed 13 self-reported questionnaires at three timepoints.The first assessment was post-Norwood surgery, at approximately 8 weeks of infant age.The second assessment was post-Stage II surgery, at approximately 6 months of infant age.The final follow-up assessment took place at a study visit at end of trial when the child was approximately 16 months of age.

Participants
Parents from 9 of the 15 Single Ventricle Reconstruction Trial (SVR) centres and parents from two additional centres, whose infants underwent staged surgical palliation for hypoplastic left or hypoplastic right heart syndrome, participated in the study.Both mothers and fathers of the same infant were eligible to participate.Parents not fluent in English were excluded.Parents whose infant died or required heart transplantation were excluded from participation, and parents were automatically withdrawn from the survey study if infant death or heart transplantation occurred during the study.Data obtained prior to infant death or heart transplantation were included in the analysis.Each centre's Institutional Review Board reviewed and approved the study.Parents provided informed consent to participate.

Measures
Questionnaires were organised into booklets and each participant was invited to complete one booklet per timepoint.Questionnaires took approximately 40minutes to complete.All questionnaires are described in detail in Supplementary Table S1.
The Impact of Event Scale -Revised was used to measure symptoms of post-traumatic stress disorder in parents. 8The 22-item self-report screening questionnaire corresponds to DSM-IV symptoms of post-traumatic stress disorder in 3 subdomains as follows: intrusion (e.g., symptoms include flashbacks, nightmares, difficulty sleeping); avoidance (e.g., symptoms include emotional numbing, altered sense of time and place, detached from self and others); and hyperarousal (e.g., symptoms include reduced concentration, irritability, and sudden onset of physical symptoms like a racing heart and trouble breathing).Data for this study were collected prior to the release of the DSM-V; however, screening tools based on DSM-IV are still widely used and reported.The questionnaire's psychometric properties have been previously established. 10Eight questions evaluated intrusion, eight evaluated avoidance, and six evaluated hyperarousal.Parents answered the questions with respect to their infant's diagnosis and ongoing clinical course.They rated each question on a 5-point Likert scale ranging from 0, "not at all" to 4, "extremely".The highest possible total score is 88, with subdomain highest scores of 32 for intrusion, 32 for avoidance, and 24 for hyperarousal.The total score provides a continuum of risk, including no risk (0-23), possible posttraumatic stress disorder (24-32), and probable post-traumatic stress disorder (> 33). 11,12,13even additional questionnaires were used to address the study aims including: Family Information Form (demographic variables); Life Stress Inventory (total burden of stressful life events); 14 State-Trait Anxiety Index (measure of transient (state) and sustained (trait) anxiety); 15 Pediatric Inventory for Parents (level of stress and appraisal of stress in parents of child with serious or chronic disease); 16 Coping Health Inventory for Parents (measures family integration and social support of parents of child with serious or chronic disease); 9 Family Inventory of Resources for Management (measures resources that impact parent self-esteem, communication, financial well-being and social support); 9 and Inventory of Parent's Experiences (measures social support network and satisfaction with parenting). 17For stress questionnaires (2, 3, and 4), lower scores reflected lower stress and for coping questionnaires (5, 6, and 7) higher scores reflected better coping.

Statistical analysis
All analyses were conducted using SAS v9.4 (SAS Institute Inc., Cary, NC, United States of America), with p-values < 0.05 representing statistical significance.Change in post-traumatic stress disorder scores over time and comparison of mothers to fathers were assessed via repeated measures regression modelling on study timepoint (as a categorical variable) and parent (mother versus father).Multi-variable, backwards stepwise repeated measures regression was used to model the relationships between post-traumatic stress disorder symptoms and other variables of interest, separated into demographic and clinical variables and stress and coping variables.Data from both parents were included when available, specifying infant as a repeated variable within the mixed model regression.All data from the same infant over time were correlated (compound symmetric variance structure).All variables with a p-value < 0.2 in bivariable modelling were included in the initial multi-variable model, and backwards selection performed.The State-Trait Anxiety state anxiety scores were included in bivariable modelling but not included in multi-variable modelling.The State-Trait Anxiety trait anxiety scores were included in both bivariable and multi-variable modelling.
Prior to model building, independent variables were investigated for substantial associations, defined as an absolute Pearson correlation > 0.5 for continuous variables, p < 0.05 on Pearson chi-square test for categorical variables, or p < 0.05 on ANOVA for the association between a continuous and a categorical variable.
No substantial associations were found among the demographic and clinical variables for the total score; a small number were found for subdomains and the variable more significantly associated with outcome on bivariable modelling was chosen to proceed to multi-variable modelling, as noted in table footnotes.Some stress and coping variables had strong correlations (Supplementary Table S8), but all except trait anxiety scores were included in multi-variable modelling as these variables are designed to measure different concepts in our conceptual framework. 9After arriving at multi-variable models for demographic/clinical variables and stress/coping variables separately, but including effects of parent and timepoint in both, all significant variables were included together in a final backwards stepwise repeated measures regression analysis to arrive at the final, combined parsimonious model.The R 2 for models (typically defined as the amount of variance in outcome explained) was calculated for these repeated measures models using the method of Jaeger et al. 18

Participant demographic information
Participants included 215 parents (143 mothers and 72 fathers) representing 146 infants (Fig 1).Demographic characteristics of parent and infants are presented in Table 1.Most infants (92%) had hypoplastic left heart syndrome, with 8% having hypoplastic right heart syndrome.Infants underwent staged surgical palliation with first surgery being the Norwood procedure in all cases.Most infants (73%) were diagnosed with CHD pre-natally.Ten percent of infants had a genetic syndrome, while 23% of infants had a nonsyndromic abnormality.
Complete Impact of Event Scale -Revised data were available for 133 parents of infants post-Norwood, 107 parents post-Stage II, and 76 parents at final follow-up (Fig 1).For the participants also in SVR, timepoint 1 occurred 53 ± 113 days after Norwood surgery, timepoint 2 occurred 56 ± 129 days after Stage II surgery or before Stage II (n = 3), and timepoint 3 occurred 348 ± 168 days after Stage II surgery.Fourteen parents at timepoint 3 representing nine children completed assessments 2-29 days after their child's Fontan procedure.Children post-Fontan procedure were 20-32 months old versus the average age of 16 months old at final follow-up.The number of participants who became ineligible over the course of the study due to infant death or heart transplantation was 22 parents.Data obtained prior to infant death or heart transplantation were included in the analysis.

Impact of Event Scale -Revised scores over time in mothers and fathers
Total parent scores on the Impact of Event Scale -Revised were highest post-Norwood and lowest at final follow-up, for both mothers and fathers (Fig 2 ; p < 0.001 for time).Each of the three subdomain scores also decreased over time (p = 0.009 for time for intrusion, p = 0.03 for time for avoidance, p < 0.001 for time for hyperarousal).Total scores were significantly lower for fathers compared to mothers across all timepoints (p < 0.001 for parent; no significant interaction with time).(IQR 11, 31) at final follow-up.Subdomain scores for intrusion were significantly lower for fathers than for mothers (p < 0.001 for parent) with a decrease over time only in mothers (p = 0.006 for the interaction).No difference between mothers and fathers was observed in the avoidance subdomain (p = 0.57 for parent), but fathers reported fewer hyperarousal symptoms compared to mothers (p < 0.001 for parent).

Prevalence of post-traumatic stress disorder in parents
The prevalence of probable post-traumatic stress disorder (total Impact of Event Scale -Revised > 33) was higher in mothers compared to fathers and decreased over time for both (Fig 3).The prevalence of probable post-traumatic stress disorder in mothers was 50% post-Norwood, 47% post-Stage II, and 27% at final follow-up, compared to 39%, 36%, and 24% of fathers, respectively.The percentage of mothers with possible post-traumatic stress disorder (total Impact of Event Scale -Revised score of 24-32) was 19% post-Norwood, 16% post-Stage II, and 19% at final follow-up, compared to 11%, 17% and 14% of fathers, respectively.
There were only 28 parents (20 mothers, 8 fathers) who participated at all three timepoints.Of these 28 parents, 25% had probable post-traumatic stress disorder at all three timepoints (5 mothers, 2 fathers), another 14% had possible post-traumatic stress disorder at all three timepoints (2 mothers, 2 fathers), and 29% had no risk for post-traumatic stress disorder at all three timepoints (5 mothers, 3 fathers).The sample size was too small for statistical testing and data are not reported for parents whose scores changed over time.

Factors associated with symptoms of post-traumatic stress disorder: Bivariable modelling
In the bivariable modelling, the only demographic variable significantly associated with symptoms of post-traumatic stress disorder was parent age (Table 2), with older parents reporting less symptoms of post-traumatic stress disorder (p = 0.04, R 2 = 0.02).Stress variables had much stronger associations with symptoms of post-traumatic stress disorder compared to demographic variables (Table 2, Supplementary Tables S2-S4) and had higher R 2 (up to

Factors associated with symptoms of post-traumatic stress disorder: Multi-variable modelling
In the multi-variable modelling, mothers (versus fathers) and earlier study timepoints (post-Norwood and post-Stage II) were significantly associated with more symptoms of post-traumatic stress disorder and more strongly associated with post-traumatic stress disorder than any of the demographic or clinical variables (Table 3, Supplementary Tables S5 and S7, R 2 = 0.16-0.20).An increased number of complications during the interstage period was the only demographic variable significantly associated with increased symptoms of avoidance (R 2 = 0.07) (Supplementary Table S6).Stress variables were significantly associated with symptoms of post-traumatic stress disorder both the total and subdomain Impact of Event Scale -Revised scores.In the multi-variable modelling, higher scores in the trait anxiety score of the State-Trait Anxiety Index indicating one's tendency towards becoming anxious along with higher scores in the Pediatric Inventory for Parents indicating more frequent and more difficulty with events parents of children with serious illness experience were both significantly associated variables of symptoms of post-traumatic stress disorder (total Impact of Event Scale -Revised score, R 2 = 0.65).Higher scores in state anxiety scores of the State-Trait Anxiety Index indicating transient situational anxiety bring the proportion of variance to 71% (data not shown, total Impact of Event Scale -Revised score, R 2 to 0.71).Trait anxiety scores versus state anxiety scores were included in the final model because trait anxiety scores are less subject to fluctuation with change in number and degree of stressful circumstances and trait anxiety scores are strongly correlated with state anxiety scores (Supplementary Table S8).Higher scores on stress scales indicating more stress and anxiety were also significantly associated variables for increased symptoms of intrusion, avoidance, and hyperarousal (Supplementary Tables S5-S7).
Coping variables were significantly associated with symptoms of avoidance and hyperarousal.Lower scores on the Family Inventory of Resources for Management, which indicated less resources to support parent self-esteem, communication, financial well-being, and social support, were significantly associated with increased symptoms of avoidance (Supplementary Table S6).Lower scores on the Inventory of Parent Experiences, signalling less social support, and reduced satisfaction with parenting were significantly associated variables for increased symptoms of hyperarousal (Supplementary Table S7).

Discussion
Symptoms of post-traumatic stress disorder are prevalent and persistent in both mothers and fathers of infants with single ventricle CHD after surgical palliation.Symptoms of post-traumatic stress disorder (total Impact of Event Scale -Revised score) are significantly associated with increased parent anxiety as well as the frequency and parent's appraisal of the degree of difficulty of events they face while caring for a child with serious illness.Reduced coping and decreased satisfaction with parenting were significantly associated with two of the subdomains (avoidance and hyperarousal) of symptoms of post-traumatic stress.
In this study, as many as 50% of mothers and 39% of fathers met the criteria for probable post-traumatic stress disorder.This is considerably higher than prevalence seen in the general population, wherein the lifetime prevalence is 8% and the 12-month prevalence is 4%. 6The disorder is associated with younger age, female gender, unemployment, single marital status, lower education, and reduced income in the general population. 6,19The prevalence of post-traumatic stress disorder in healthy post-partum mothers, with no pre-existing mental health conditions, is 5%-13%. 20,21,22,23Symptoms steadily decline around 6 weeks postpartum in healthy mothers. 20Data on post-traumatic stress disorder in healthy fathers in the post-partum period are very limited although one report found no significant symptoms across the three subdomains while another reported 5% of fathers had symptoms of post-traumatic stress disorder. 24,25Higher prevalence ) 19 -24 % of mothers and 13% of fathers) of post-traumatic stress disorder is observed in parents of preterm infants and infants with fetal abnormality. 26,27,28The prevalence of post-traumatic stress disorder in parents of children with single ventricle CHD post-Norwood and post-Stage II palliation is comparable to that seen in bereaved mothers (42-69% of bereaved mothers). 29,30f parents of children who require bypass surgery for mild to severe CHD, 25% of parents met criteria for post-traumatic stress disorder 1-month post-surgery and 15% of mothers and 10% of fathers still met criteria for post-traumatic stress disorder at 6 months post-surgery. 31,32Data on parents further out from their child's surgery are lacking, but in a single centre study with 29 parents (16 mothers, 13 fathers) of children with hypoplastic left heart syndrome who were interviewed one time when their child was age 1-18 years old, 88% of mothers interviewed had symptoms of post-traumatic stress disorder and 66% of fathers had symptoms of post-traumatic stress disorder. 3,33,34It is not clear if parent's symptoms of post-traumatic stress disorder fully resolve or whether post-traumatic stress disorder symptoms remain prevalent months to years after their child's surgery for single ventricle CHD.Continued research is needed.
Post-traumatic stress disorder has a high level of comorbidity with other mood disorders, anxiety disorders, disruptive behaviour disorders, chronic physical conditions, and addictions. 6,35Poorer parent mental health including higher parent stress and anxiety has been found to be associated with poorer neurodevelopmental outcomes in children with CHD. 36Parental post-traumatic stress symptoms severity is associated with overprotective parenting as well as emotional and behavioural problems in children who have CHD. 5 Parental post-traumatic stress is also associated with an increase in unplanned hospital visits in the first few months of their infant's life. 37Family-based psychosocial interventions that empower parents as caregivers and advocates for their children with CHD could reduce the impact of parent mental health problems and improve child outcomes. 38,39,40Effective treatments provided early on for post-traumatic stress disorder such as cognitive-behavioural therapy have been tested in many settings showing effective results. 41Our results demonstrate that mothers and fathers could benefit from early identification through screening for anxiety during hospitalization with continued monitoring for development of symptoms of post-traumatic stress especially symptoms that persist longer than 30 days after cardiac surgery as persistence of symptoms is part of criteria for diagnosis.
It is important to consider the reality that single ventricle CHD carries the highest morbidity of any form of CHD with a significant risk of death or heart transplantation. 42All single ventricle CHD children require additional cardiac surgeries, catheter procedures, and a significant number will have complications including pacemaker placement, thrombotic events, stroke, seizure, protein- Cardiology in the Young losing enteropathy, and plastic bronchitis. 42Complications can occur at any point in the child's life and their parents likely repeatedly experience trauma over time as their child requires further palliation that necessitates intensive care support, lengthy hospitalizations, and uncertain futures. 42Screening for symptoms of posttraumatic stress disorder should be on-going pre-operatively through hospitalization and again post-operatively.Early identification, appropriate resource allocation to those parents who are deemed at-risk, empathic communication, and consistency in care could help support parents. 2,4,43ur study showed that demographic and clinical (medical or surgical) factors such as pre-natal diagnosis, complications, and length of hospital stay did not identify which parents were at greater risk to develop post-traumatic stress disorder symptoms, a finding supported by other studies. 44,45Measures of anxiety have been associated with elevated post-traumatic stress disorder symptoms in parents of children with CHD. 45,46Like our findings, Turgoose et al. did not find significant demographic or clinical factors that predict post-traumatic stress disorder in parents of children with CHD (any type) undergoing bypass. 45Parents need support and education to help them adapt to the stress and challenge of raising a child with critical CHD. 47,48Healthcare professionals are in a unique position to have a positive influence on how parents adapt to stressors by recognising symptoms in parents and referring parents to appropriate resources.
Study participant withdrawal was a limitation of this study which prevented analysis of paired parent dyads.There were too few parents with longitudinal data to address the question of change in probable post-traumatic stress disorder between timepoints per parent.The most common reason for participant withdrawal from the study was parent stress.Parents who were struggling the most may not have participated at some timepoints or may never have consented at all.It is also a possibility that survey fatigue due to the time commitment required to complete the questionnaires contributed to attrition rates.The Family Adaptation study was designed to capture the elements of a complex psychosocial process and surveys were selected based on the framework of stress appraisal, coping, and adaptation; however, all surveys contributed to the framework and allowed thorough analysis of findings.This study relied on self-report measures and the Impact of Event Scale -Revised is a screening tool versus formal diagnostic tool for post-traumatic stress disorder.The high prevalence of probable post-traumatic syndrome disorder on screening does suggest that future studies should consider a formal diagnosis through clinical standard diagnostic interview to better understand the true prevalence in this vulnerable parent population.
Attention to parent mental health and overall well-being during uncertain and stressful times is important.These results suggest more help is needed for parents, and screening and intervention might be beneficial.
Supplementary material.For supplementary material accompanying this paper visit https://doi.org/10.1017/S1047951122004012Table 3. Multi-variable repeated measures regression modelling for IES-R total score versus demographic variables and versus stress and coping variables.The final combined model was the same as the model for stress and coping variables.a.The total n includes all participating parents across all timepoints.The statistical model used, repeated measures regression, accounts for this by treating all data entries for the same infant as correlated with each other, rather than as independent information

Figure 1 .
Figure 1.Consort diagram of study participant overall enrolment from trial centres and number of study participants across 3 timepoints.HLHS=hypoplastic left heart syndrome; PHN=Pediatric Heart Network; SVR=Single Ventricle Reconstruction; IES-R=Impact of Event Scale -Revised.

0
.52 for the total Impact of Event Scale -Revised score).Higher scores of stress variables (indicating increased stress and anxiety) in the Inventory of Life Changes, State-Trait Anxiety Index, or Pediatric Inventory for Parents were significantly associated with symptoms of post-traumatic stress disorder, while lower scores of coping variables (indicating reduced coping and social support) in the Inventory of Parents Experiences or Family Inventory of Resources for Management were significantly associated with symptoms of post-traumatic stress disorder.

Figure 2 .
Figure 2. Median total and subdomain IES-R scores by timepoint by parent: (A) total; (B) intrusion; (C) avoidance; (D) hyperarousal.Post-Norwood ~8 weeks infant age.Post-Stage II ~6 months infant age.Final follow-up ~16 months child age.IES-R, Impact of Event Scale -Revised.Results of repeated measures regression demonstrate the total Impact of Event Scale -Revised scores were highest post-Norwood and lowest at final follow-up, for both mothers and fathers (p < 0.001 for time).Total scores were significantly lower for fathers compared to mothers across all timepoints (p < 0.001 for parent ; no significant interaction with time).A total Impact of Event Scale -Revised score greater than 33 indicates probable post-traumatic stress disorder.Subdomain scores also decreased over time (p = 0.009 for time for intrusion, p = 0.03 for time for avoidance, p < 0.001 for time for hyperarousal), though symptoms of intrusion only decreased over time for mothers (p = 0.006 for the interaction).

Figure 3 .
Figure 3. Stacked bar plots of IES-R scores as percentages by timepoint, stratified by parent.50% of mothers and 39% of fathers have probable post-traumatic stress disorder post-Norwood versus 27% of mothers and 24% of fathers at the final follow-up when child is approximately 16 months old.Data presented for complete IES-R questionnaires.Missing data due to incomplete IES-R questionnaire, non-participation, and ineligibility excluded.Post-Norwood ~8 weeks infant age.Post-Stage II ~6 months infant age.Final follow-up ~16 months child age.PTSD=post-traumatic stress disorder; IES-R=Impact of Event Scale -Revised.

Table 1 .
Demographic characteristics of infants, mothers, and fathers *Mothers and fathers could respond to demographic questions regarding the other parent, without the second parent being represented in the survey responses.Mean ± sd for continuous variables and % for categorical variables.Percentages may not add up to 100% due to rounding.Parent education was a proxy for socio-economic status.

Table 2 .
Bivariable repeated measures regression modelling for IES-R total score p < 0.05, **p < 0.01, ***p < 0.001.Total n includes a category for "unknown", which is not presented.The total n includes all participating parents across all timepoints.The statistical model used, repeated measures regression, accounts for this by treating all data entries for the same infant as correlated with each other, rather than as independent information.LOS=length of stay; IES-R=Impact of Event Scale -Revised. *