Parental traumatic stress during and after paediatric cancer treatment

Abstract

The objective was to cross-sectionally compare parents of children during (n = 175) and after (n = 238) cancer treatment regarding traumatic stress (intrusion, avoidance, arousal). In both groups, time since child's diagnosis ranged from one month to six years. Intrusion and arousal were more frequent in parents during ongoing treatment, although also reported by many parents after treatment. Stress was evaluated in relation to situational and demographic factors: Parents who had experienced a relapse did not differ from parents of non-relapsed children. Time since diagnosis was only weakly associated with stress. In the stage of completed treatment the risk for severe stress was elevated in parents with lower education and immigrant parents. Mothers reported somewhat higher levels of stress than fathers, although the findings were ambiguous. To conclude, many parents experience high levels of disease-related stress, even after successful treatment. The pattern of stress symptoms may vary according to educational level, ethnicity, and gender.

Cancer in a child radically alters the life situation of the parents and brings about a range of stressors of varying impact. The traumatic potential of the disclosure of the cancer diagnosis and the realization of the fatality of the disease has been confirmed by research during the past years [1,2]. In addition to concerns about the child's health, parental stressors relate to, for example, communicating with siblings of the ill child, accepting the intensity of one's own reactions, and dealing with the reactions of others [3]; concerns about the child's future [4]; negative employment and financial consequences [3]; quality of care [5]; and aspects of the cancer treatment itself [6,7].

During the treatment phase, many of the triggers eliciting distress in parents may be external and situational. Conversely, when treatment is completed the stressors may rather be internal, such as worries concerning the development of the child, fear of a relapse, and memories of traumatic experiences during the course of the disease. The latter may yield stress reactions even years after a successfully completed treatment of the child [e.g. 8,9].

In research, parental reactions to their child's cancer have typically been assessed in terms of psychopathology, e.g. posttraumatic stress disorder (PTSD), or subclinical psychiatric symptomatology, such as depression, anxiety, and general psychiatric symptoms. Although the stress caused by the child's illness only in a minority of the cases seems to result in psychopathology, parents may be bothered by strong stress reactions during ongoing treatment, as well as after treatment [10–12]. Although these stress reactions stay in the non-pathologic range, such long-term exposure to stress may have subjective and objective health consequences [13,14], as well as negatively influence the social support network [15]. Paying attention to the non-pathological stress reactions therefore appears to be an appropriate complement to screening for psychopathology in parents of children with cancer.

When studying immediate as well as long-term responses to a severe stressor, reactions are commonly categorized in three types: intrusive thoughts of the stressor, avoidance of distressing reminders, and increased arousal. In research on parents of childhood cancer, a composite score of the three response types has typically been used, while the distribution of symptoms within the separate categories has only occasionally been reported. Most studies have covered parents whose children have completed cancer treatment [8], [9], [16–19], but researchers have also reported on parental posttraumatic stress symptoms 1–2 months after the first hospitalisation of the child [2]. Manne and colleagues studied 65 mothers after the end of their child's cancer treatment, and found that 6.2% met the criteria for PTSD [16]. In two other studies, also post-treatment and restricted to mothers, 25% of the studied mothers were diagnosed with current PTSD [8,18].

Although similar assessment measures have been used, the conceptual terminology varies, and includes posttraumatic stress, posttraumatic stress disorder, posttraumatic stress symptoms, and traumatic stress. Concepts have differed, depending on whether focus of interest has been in the incidence of psychopathology, in terms of meeting the diagnostic criteria for PTSD, or whether the interest has been in assessing stress of a traumatic origin, regardless of whether a psychiatric diagnosis can be established or not. In this study, belonging to the second of these two study categories, the term traumatic stress is used to comprehensively cover trauma-induced stress [20]. Traumatic stress is here defined as symptoms of intrusion, avoidance and arousal in relation to experiences in the past as well as recent experiences, and consequently not restricted to symptoms of posttraumatic stress.

Despite the distressing and long-lasting character of the child's treatment, parental traumatic stress throughout this phase has to our knowledge not yet been studied. Thus, the main aim of this study was to investigate traumatic stress in parents of children in active cancer treatment, compared with parents after the end of treatment. A further aim was to investigate any relations between traumatic stress on one hand, and five either disease related (relapse, time elapsed since diagnosis) or demographic (educational level, ethnicity, and gender) factors on the other hand.

Method

Participants and procedures

Four hundred and twenty three parents of children with ongoing (103 mothers, 78 fathers) or completed (134 mothers, 108 fathers) curative cancer treatment were assessed on one single occasion between October 2000 and April 2003. Parents were recruited at the paediatric oncology units of two Swedish hospitals, Astrid Lindgren Children's Hospital and Linköping University Hospital, through inviting all parents who had a contact at the unit at a certain period of time. Non-Swedish speaking parents were not included. The invitation of parents was done by one of the researchers or by a research nurse, either personally while parents were visiting the hospital, or by phone or mail. Along with the questionnaires parents were provided with written information about the project. They were instructed to complete the questionnaires independently of each other. After completing the questionnaires at home, parents returned them by mail in a pre-paid return envelope. The overall response rate was 73%. Characteristics of the study group are presented in Table I. Regarding time elapsed since the child's diagnosis, the in-treatment and off-treatment groups were equivalent in as much as they ranged from 1 to 74, and 1 to 72 months post-diagnosis respectively. However, the in-treatment group was somewhat biased, with most of the parents assessed early during treatment. In the in-treatment group, 77% were assessed during the first year, 11% during the second year, 5% during the third year, and 7% during a period from the fourth year to 6 years 2 months after diagnosis. The corresponding figures for the off-treatment group were 19% during the first year, 16% during the second year, 28% during the third year, and 37% from the fourth year to 6 years 2 months. The data analysed in this study were collected as part of a larger ongoing study, investigating the psychological and psychosocial situation of parents of children with cancer. The study was approved by the local ethics’ committee.

Table I.  Characteristics of the study group.

	Parents of children in treatment,% (n)	Parents of children off treatment,% (n)
Mothers	57 (99)	55 (131)
Fathers	43 (76)	45 (107)
Parent's education:
Elementary school only	10 (17)	17 (41)
College/high school	46 (80)	47 (111)
University	44 (77)	35 (84)
Missing data	0.6 (1)	0.8 (2)
Born in Sweden	79 (139)	83 (198)
Immigrant to Sweden	21 (36)	17 (40)
Child had not suffered a relapse	92 (161)	90 (215)
Child was in treatment for relapse	8 (14)	n.a.
Child had completed successful treatment for relapse	n.a.	10 (23)
	mean (SD); range	mean (SD); range
Age of child at assessment, years	7 (5); 0.1/ 20	10 (6); 0.2/ 23
Time from child's cancer diagnosis to assessment, months	10 (15); 1/ 74	32 (18); 1/ 72

n.a.: not applicable

Measures

A revised version (IES-R) of the Impact of Event Scale (IES) was used for assessing stress reactions [21,22]. The original IES and the expanded IES-R are two of the most widely used self-report instruments in trauma research and have proven to be useful in the assessment of psychological stress after traumatic events [e.g. 23]. The three aspects of stress reactions covered by the instrument, intrusion (8 items), avoidance (8 items) and arousal (6 items), are congruent with the B, C, and D symptom categories of the diagnostic criteria of posttraumatic stress disorder [24]. Although originally developed to measure reactions to current stressors, the scale is frequently used when the purpose is to assess symptoms of posttraumatic stress. The instrument measures the extent to which respondents have experienced the symptoms during the last seven days. The five response alternatives “Not at all”, “A little”, “Moderately”, “Quite a bit”, and “Extremely” are scored 0 to 4. Individual results are expressed by sum scores, with higher scores indicating the presence of more symptoms. According to the instructions, parents were told to answer the questionnaire with reference to the illness of their child; i.e. they could report symptoms, related to any aspect of the child's cancer. Previous studies present evidence suggesting that the English version of the IES-R has adequate reliability and validity, although the psychometric properties of the arousal subscale have been somewhat weak [23,25]. In the present study, Cronbach's alpha coefficients indicated satisfactory to good internal consistency for all three subscales: intrusion .90 (in treatment) and .91 (off treatment); avoidance .77 (in treatment), and .86 (off treatment); and arousal .83 (in treatment), and .85 (off treatment).

Information about relapse and time elapsed since diagnosis was obtained from the medical records. Data on the three demographic background variables were collected through the questionnaire: parent's educational level (elementary school only, college/high school, or university); ethnicity (Swedish-born or immigrant to Sweden); and gender.

Data management and statistical analyses

Ten questionnaires in which the respondents had left more than 25% of the items unanswered on one or more of the IES-R subscales were excluded from the analyses, leaving a study group of 413 parents for which data were analysed. For parents with 25% or less of the items unanswered, missing values were replaced by the individual mean score of the subscale in question. Such imputations were made for 16 answers to items in the intrusion subscale, and for 13 and 5 answers to items in the avoidance and arousal subscales, respectively. Respondents with data missing in a covariate or background variable were excluded from the analyses of that particular variable. The analyses were performed separately for parents whose children were still in treatment, and for those of children who had completed treatment.

Comparisons of parents of children in treatment and parents of children off treatment, with regards to IES-R sub-scale scores were done using unpaired t-test. The IES-R scores were additionally examined by computing frequencies of parents reporting symptoms. Parents who had responded either with “Quite a bit” or “Extremely” to at least one item in a subscale were considered as presenting symptoms, while the answers “Not at all”, “A little”, and “Moderately” were regarded as reflecting an absence of symptoms. In order to investigate the severity of traumatic stress, we also examined the amount of parents presenting a profile of symptoms, resembling the clinical representations of traumatic stress, defined in the DSM-IV diagnostic criteria of PTSD [24]: respondents who reported at least one symptom mentioned in the intrusion subscale, at least three symptoms of avoidance, and at least two symptoms of arousal were identified. Parents exhibiting this symptom profile were in this study considered as demonstrating “clinically significant traumatic stress”.

To examine possible differences relating to the background variables, we executed a series of separate bivariate analyses. Traumatic stress in relation to parents’ experience of a relapse was investigated using unpaired t-tests. Parents of children in treatment who had never suffered a relapse were compared with parents of children who were in treatment for a relapse. In the same manner parents whose children had completed treatment for a relapse were compared with parents whose children had completed treatment without a relapse. The variation in traumatic stress symptoms in relation to time elapsed since the child's diagnosis was examined with Pearson correlation analyses (2-tailed). Educational levels were examined by one-way ANOVA, and differences in level of stress in relation to ethnicity were analysed with unpaired t-test. Possible sex differences in stress reactions were analysed through t-tests. To avoid the potential bias of dependent observations, caused by data emanating from mothers and fathers of the same family, two different series of analyses were done: (a) the scores of the 20 fathers and 69 mothers from families where only one parent had participated were compared by unpaired samples t-tests, and (b) paired samples t-tests were used to compare levels of intrusion, avoidance, and arousal in mother and father from the 162 couples where both parents had participated in the study.

Because of the total number of statistical tests performed, an adjusted alpha level was applied to fend off the risk of Type I error. Thus, a p-value of <0.01 was regarded as statistically significant.

Results

Total scores for reported symptoms are presented in Table II. Parents of children in treatment reported significantly more intrusive thoughts (t = 4.7; p < 0.001) and arousal (t = 4.7; p < 0.001) than parents of children off treatment, while avoidant behaviours were equally reported in both groups.

Table II.  Comparison of traumatic stress in parents of children in cancer treatment and parents of children off cancer treatment.

	Parents of children in treatment (n = 175)	Parents of children off treatment (n = 238)
Intrusion, mean (SD)	13.2 (7.6)***	9.7 (7.4)
Avoidance, mean (SD)	6.0 (5.1) ^n.s.	5.4 (5.9)
Arousal, mean (SD)	7.6 (5.8)***	5.0 (5.1)
Intrusion, “quite a bit” or “extremely”	71%	59%
Avoidance, “quite a bit” or “extremely”	37%	28%
Arousal, “quite a bit” or “extremely”	54%	29%

*** p<.001; n.s.: not significant

Intrusion was the most frequently reported symptom, and was “quite a bit” or “extremely” bothering seven out of ten parents whose children were still in treatment, and more than half of the parents of children off treatment (Table II).

Of the parents of children in treatment, 23% met the criteria for clinically significant traumatic stress as defined here, by exhibiting at least one symptom of intrusion, three symptoms of avoidance, and two symptoms of arousal. The corresponding proportion of parents of children off treatment was 17%. The difference between the two groups, regarding incidence of clinically significant traumatic stress, was not statistically significant (Chi-square 2.80; p = 0.094).

Parents whose child had suffered a relapse reported traumatic stress at levels equal to those of parents of non-relapsed children. This applied to parents during treatment as well as parents assessed after completed treatment. Time elapsed since diagnosis was weakly negatively related to intrusion among parents of children off treatment (r = − 0.17, p = 0.008).

Among parents of children in treatment no group differences were found regarding education and ethnicity. However, among parents of children off treatment, several relations were found. Parents with a shorter education generally reported more intrusion (F = 8.8, p < 0.001), avoidance (F = 5.3, p = 0.006), and arousal (F = 10.6, p < 0.001), compared with parents of higher educational status. Parents of a non-Swedish origin generally reported more intrusion, (t = 4.0, p < 0.001), avoidance (marginally significant) (t = 2.6, p = 0.011), and arousal (t = 3.4, p = 0.001) than parents born in Sweden.

None of the three symptom clusters differentiated between the groups of unrelated mothers and fathers. In contrast, paired samples t-test of the average differences within couples indicated that the mother was the one in the parent couple usually reporting a higher level of intrusion (in treatment marginally significant, t = 2.5, p = 0.015; off treatment, t = 3.3, p = 0.001). Avoidance and arousal were reported equally.

Discussion

Traumatic stress was more common among parents of children in treatment than in parents of children off treatment. Non-Swedish origin and low education increased the risk for parental stress symptoms post treatment.

The higher levels of stress symptoms reported by parents of children in treatment compared with parents after treatment end is most reasonably a consequence of the exposure to a broader array of potentially stressful disease-related events during treatment. According to our clinical experience, uncertainty about the success of treatment stands out as a particular source of worry. Any expressions of posttraumatic stress during this phase may indeed be accompanied by reactions to current stressors. Nevertheless, fear of a relapse, and worry about evolving negative consequences of the treatment may also be intensified during the period following the end of treatment. Accordingly, current stressors related to the child's disease may be present also after completion of the treatment.

It is a difficult task to compare reported stress levels in different traumatic situations and such comparisons should be done with caution. Still, such comparisons approach a rough estimation of the traumatic impact of different circumstances. In the present study, both groups of parents exhibited higher scores of intrusion, avoidance, and arousal, than what has been reported in a study involving patients with the experience of a life-threatening cardiac event [25], but lower than those reported in a study on traumatized refugees with the clinical diagnosis of PTSD [26]. Along the same line, comparisons could be made between incidence of PTSD in other studies, our concept “clinically significant traumatic stress” (23%in, 17% off treatment) and other expressions of more severe traumatic stress. For instance, Landolt and colleagues [2] reported a 71% incidence of PTSD in a group of 19 parents, whose children were recently diagnosed with cancer (n = 7) or had been involved in a traffic accident (n = 12). Other studies, not specifically designed to assess the incidence of clinical PTSD, have reported symptoms in the areas of intrusion and avoidance, and sometimes also hyperarousal. In most of these studies, a total score has been used, whereas symptom categories have not been evaluated separately. The incidence of severe posttraumatic stress symptomatology among parents after the end of treatment has been reported to be between about 10 and 40%: Stuber, Christakis, Houskamp and Kazak [17] concluded that 39% mothers and 33% fathers in their sample exhibited severe posttraumatic stress symptomatology, while the corresponding numbers reported by Kazak et al. [19] were about 10%, and in a study by Barakat et al. [9] 10% and 7%.

For the clinician, it is important to understand whether the suffering for which he or she is supposed to support a parent is primarily caused by the current situation, or whether it relates to past experiences during the course of the illness. Parents’ struggling to cope with present stressors as well as those showing signs of persistent posttraumatic symptoms may both be in need of professional support, although the ways of meeting their needs may differ regarding type and duration of support interventions. Parents reacting to the acute stresses may benefit enough by being seen and given the opportunity to share their experiences, and, when needed, being assisted with support and tools to handle a temporary strenuous situation. When persistent posttraumatic symptoms remain, however, the professional effort may have to cover a more elaborated treatment approach. This includes multimodal treatment where attention is paid to the symptoms of intrusive thoughts and re-experiencing, and avoidance, as well as to the unspecific symptoms following PTS [27,28]. Referral to a team possessing the resources and treatment expertise needed may be required in those cases.

The absence of a relationship between parental stress and relapse is to some extent contrary to previous findings. A relapse has previously been found to predict psychological distress, and to have the most impact for mothers [29]. Concerning fathers, earlier findings are more ambiguous. For instance, in two Asian studies, the psychological distress of fathers of children with cancer was not at all affected by a relapse [30,31]. One possible interpretation of our results is that for many parents the cancer diagnosis may emotionally already be linked to a risk of a relapse, regardless of the initial prognosis in the particular case. However, the restricted number of parents of relapsed patients in the present study limits the generalizability of the findings regarding this subgroup. As it seems, parents’ experiences of a relapse of childhood cancer warrant further investigation.

Intrusive thoughts and images associated with the child's disease represented the most frequently reported symptom type in both groups. This is in line with previous findings of North American studies of parents after the end of treatment [16,17]. Symptoms of intrusion were slightly more frequent closer in time after diagnosis. However, the weakness of the associations indicate that the systematic variation of traumatic stress symptoms as regards time elapsed since the child's diagnosis is almost ignorable, and symptoms may be present in parents at any point in time during the first six years following a child's cancer diagnosis. Psychiatric symptoms and anxiety, in contrast, have been found to decrease during the first years of treatment [32,33]. This points to that the assessment of context-related reactions such as traumatic stress may be a more sensitive measure than evaluation of psychopathology or negative emotions, when investigating the psychological impact of childhood cancer on parents.

Among the parents of children off treatment, those with lower education and immigrant parents reported higher levels of stress. We believe that these results – on a group level – reflect differences in vulnerability and situational adaptation. Parents with lower education as well as immigrant parents may be less resilient to stressors and have poorer prospects of recovery after a period of severe stress, as they often are exposed to concurrent psychosocial stressors and subject to other unfavourable conditions [34,35]. During ongoing treatment, on the other hand, the level of stress symptoms did not differ according to these demographic variables. The variation in level of stress seems to be outrun by the higher distress associated with unfinished treatment. Sociodemographic factors such as education and ethnicity are not supposed to have a direct influence on stress or mental well-being, but rather to include other underlying aspects relating to social, economic and material strain, which in turn may affect factors such as self-esteem, locus of control, and stressor appraisal. As regards to immigrants, previous and concurrent experiences of discrimination, violence, and disrupted social networks may also influence stress resiliency.

Although mothers and fathers in Sweden today increasingly share the practical responsibilities of the family equally, the mother still appears to be the one who spends more time with the child at the hospital. As a consequence mothers may be more exposed to the stressors associated with caring for the sick child, and may also to a greater extent carry memories of traumatic experiences. Thus, it is not surprising that within the parent couples in the present study, the mother was generally the one reporting more intrusive thoughts. At the same time, the absence of differences on a group level, i.e. between unrelated mothers and fathers, complicates the interpretation of gender effects. A liable explanation to the inconsistency regarding these findings is that gender differences were too small to be detected by the unpaired samples test. The indication that the experience of the child's illness often is as stressful for fathers as for mothers highlights the necessity of paying attention to both parents in paediatric cancer care.

Some methodological limitations of this study need to be recognized. Firstly, the parents were assessed cross-sectionally, and, consequently, assumptions about individual processes over time cannot be drawn. Secondly, the in-treatment parent sample in this study was somewhat skewed with respect to time elapsed since diagnosis, with most of the parents assessed during the first year of treatment. This skewness may have affected the evaluation of variation in stress in relation to time. It may also partially explain the higher mean level of stress in the in-treatment group, and may consequently have biased the comparison of stress levels in the in-treatment and off-treatment groups. Finally, in this study one single measure of traumatic stress has been used. The use of additional instruments would have provided a broader and, perhaps, more reliable basis for conclusions.

Findings highlight the need for continued attention to parent disease-related stress at least during the first six years after the diagnosis, regardless whether the child's treatment is ongoing or completed. Furthermore, findings regarding parents of children in active treatment, and parents of children for whom treatment is completed demonstrate the value of considering context-related stress reactions in research and in clinical work with parents of children with cancer, in addition to screening for psychopathology. Forthcoming studies of mother-father divergences, the situation of immigrant families, and parents with lower socioeconomic and educational status may benefit from more complex approaches. These could cover the study of possible effects of interaction between sociodemographic factors, as a complement to the descriptive comparison of stress.

We gratefully acknowledge the time and energy contributed by all parents during the progress of this study. The study was supported by The Children's Cancer Foundation of Sweden, and Cancer- och Trafikskadades Riksförbund.