The psychological and existential challenge of parenting a child who has cancer influences the lives of mothers and fathers for years, even after the end of treatment.1-3 Among the resources and individual resilience factors, the availability of social support seems to be essential to the maintenance of psychological and physical well-being among parents of children with cancer.4-8 Generally, in the face of severe stress, social support is supposed to work beneficially in several ways: by affecting the use of various coping strategies,9,10 improving self-efficacy,11 and influencing the appraisal of the stressor.9
In a general theoretical model, as suggested by, for example, Hobfoll and Vaux,12 3 aspects of the social support construct can be distinguished: network resources, behaviors, and subjective appraisal of support. Network resources refer to the people available for supportive interaction (number of persons). The behavioral aspect refers to seeking support and accepting offered support. The subjective view, a belief that others are willing to provide assistance, if needed, is usually signified by the term perceived social support.13 This aspect of support can be described as part of a person's existential perspective. On examination, the frequency of social behaviors and the number of people available for social interaction may predict psychological well-being. However, subjectively perceived support seems to be the factor most strongly associated with well-being in various contexts.8,12,14
Researchers have assumed a permanence in the experience of perceived support because it has been conceptualized as a relatively stable schema.13 Nonetheless, the experience of severely stressful events may have a deteriorating effect on perceived support, as has been described by Norris and Kaniasty,14 when studying the effects of exposure to natural disaster. This deterioration seems to be relatively lasting, but receiving sufficient support has been supposed to have a protecting effect on perceived support.14 Indeed, there seems to be an initial mobilization of the support network at the time of a traumatic and/or psychologically threatening event. However, in the case of long-lasting stress, the available support recedes as time passes because chronic stressors seem to influence the social support network in a negative way.15,16 This initial mobilization and subsequent decline of supportive interactions have also been seen among parents of children with cancer.17 At the same time, in some individuals, a severely stressful event may, in the long run, lead to growth and positive change and a higher satisfaction with social support.18 Thus, perceived support may deteriorate or improve in the aftermath of severe stress.
The association between stressful events, psychological distress, and social support is complex, and the patterns of influence may be reciprocal. As indicated above,4-6,8 support may buffer against emotional distress. Also, depressive symptoms, especially when combined with low anxiety, generate a tendency to seek isolation.19,20 Furthermore, even if support is offered, the cognitions associated with depression include a tendency not to perceive support as available.21,22
Through our clinical contacts with parents, we got the impression that some of them had developed a stronger faith in life while others had grown more disillusioned after the experience of a cancer illness in their child. In a recent preliminary study, no difference was found in the mean scores of perceived support between parents of children with cancer and reference parents of healthy children.23 However, the variation of the scores was significantly larger in the study group than in the reference group. Consequently, parents of children with cancer more often reported good and poor perceived support at the extreme points of the assessment scale. Together, these clinical observations and the previous research regarding other severely stressful events indicated the possibility that parents' experience of childhood cancer initiates either one of two trajectories with regard to perceived support: a change to the better, stronger, and more positive or a change to the worse. This subject matter is important because resource loss and gain seem to be critical mechanisms in the entire stress process.24 Nonetheless, the ability of cancer in a child to influence perceived support among parents has, to our knowledge, not yet been studied.
Study Aim
We wanted to examine whether perceived support is influenced also in the context of parenting a child with cancer. Thus, the aim of the present study was to prospectively investigate perceived social support among parents of children with cancer. Two research questions were posed: (1) Is parents' perceived support altered during the period from the beginning of the child's treatment to after completion of treatment? and (2) Is it possible to predict the direction of any change in perceived support by considering initial depression or anxiety or parental gender?
Methods
Design
Longitudinal data were analyzed in the study: each parent had completed questionnaires on 2 occasions. All parents had children in the first 6 months of curative cancer treatment on the first occasion (T1) and with completed successful treatment on the second occasion of assessment (T2), 12-24 months later. This time frame was chosen to be sufficient for detecting any change in the variable under study. Data collection was carried out between February 1999 and January 2005. The study was approved by the local ethics committee.
Participants
The study sample was composed of 51 parents (29 mothers and 22 fathers) of children diagnosed with malignant disease, recruited at Astrid Lindgren Children's Hospital, Stockholm, Sweden.
The data for this study were collected as part of the entirety of records of a larger study investigating the psychosocial situation of parents of children with cancer (see Boman et al3 and Lindahl Norberg et al8,25,26). The parents included in this larger study had an ongoing contact either with the inpatient unit or the outpatient follow-up clinic. Part of the parent group had recently been told of the child's diagnosis and were consecutively invited at their initial contact with the unit. Another part was recruited among parents whose children were in a later treatment stage or in off-treatment follow-up. These parents were selected by inviting all parents visiting the follow-up clinic during 2 randomly selected months. Two exclusion criteria were applied: insufficient knowledge of the Swedish language to comprehend the questionnaires (parents who required an interpreter when communicating with the medical staff) and that the child had a known fatal malignancy (parents of children for whom curative treatment was discontinued). A follow-up design was applied for this larger study, including parents for whom various time periods had elapsed since the child's diagnosis for an initial assessment (T1) and a follow-up assessment (T2) approximately 12 to 24 months later. Three hundred forty-nine parents of children in or off cancer treatment were initially invited, among which the overall response rate at the first assessment was 77%. Of those who were eligible for a second assessment (child not in palliative treatment or deceased and family still living in the catchment area), the response rate at T2 was 82%.
A subgroup was selected from this entire group of parents in the larger study to form the present study group: parents for whom the T1 assessment was done within the first 6 months of the child's cancer treatment; and T2, after completed treatment. These selection criteria were applied to make the sample homogenous regarding time elapsed since the child's diagnosis. Of the 52 parents fulfilling these criteria, 1 respondent who had left more than 25% of the items of the perceived-support scale unanswered was excluded from the analyses. The remaining 51 parents thus constituted those for whom data were analyzed and are presented in this study.
At T1, the time elapsed since the child's diagnosis ranged from 1 to 6 months in the group. The time elapsed between T1 and follow-up assessment at T2 was 12 to 24 months. Ten were parents of children with leukemia, whereas the corresponding numbers were for lymphoma were 7; central nervous system tumor, 4; neuroblastoma, 6; Wilms tumor, 6; bone tumor, 9; soft-tissue sarcoma, 5; and germ-cell neoplasm, 4 parents.
Data Collection
The data analyzed in this study were a subsample of the entirety of records of a larger study. Parents were invited to participate in this larger study by one of the authors or by a research nurse, either personally, while parents were visiting the hospital, or by telephone or mail. Along with the questionnaires, written information about the project was also provided to the parents. They were instructed to complete the questionnaires independently of each other. After completing the questionnaires at home, parents returned them by mail in a prepaid return envelope.
Assessment Instruments
PERCEIVED SUPPORT
Parents' perceptions of social support were assessed using a self-report scale designed by Van Dongen-Melman and her colleagues as part of an extensive questionnaire for assessing the psychosocial situation of parents of children with cancer.27,28 The scale contains 8 items, 5 of which concern general social relationships and 3 that are context related concerning the child's disease (Table 1). Respondents rate the items on a 4-point Likert scale: "I strongly agree," "I agree," "I disagree," or "I strongly disagree." The individual is assigned a measure of perceived social support by dividing the total individual score by the number of nonmissing items, thus obtaining a mean value within the range 1 to 4. A higher value reflects more satisfaction with support. Cronbach [alpha] value for the scale was .84 in the Dutch study28 and .85 and .89 at T1 and T2, respectively, in this study, indicating satisfactory reliability. Evaluation of the validity of the scale has not been done.
Originally, the scale was assigned inverse scores because the scale was intended to capture feelings of loneliness, although several items address issues directly relating to the experience of support from others. However, the concept of loneliness is proposed to include both feelings that one's social needs have gone unmet and lack of confidence in the availability of others to meet these needs in the future.29 Furthermore, according to Pierce et al,29 general perceptions of support availability are more strongly associated with loneliness than are relationship-specific perceptions of available support. Moreover, the concept of perceived social support is not specific to a particular relationship or type of support (emotional support, advice, or tangible aid).13 Consequently, the scale used here arguably covers the concept even better than instruments relating to particular individuals within the social environment of the subject.
ANXIETY AND DEPRESSION
Anxiety and depression were assessed with self-report questionnaires. A 17-item version of the state anxiety scale from Spielberger's State-Trait Anxiety Scale was used as a measure of anxiety. Depressive symptoms were assessed by a 10-item scale derived from the Zung Self-Rating Depression Scale. The remaining items of the scales kept their original wording. The scales were abridged as part of the construction of an extensive multidimensional assessment tool for use with parents of children with cancer, done by Van Dongen-Melman and colleagues.27,28 The response format of both scales complies with a 4-alternative Likert scale, and the individual mean scores constitute the outcome measure. Higher scores reflect higher levels of anxiety or depression. Cronbach [alpha] values for the anxiety and depression scales were .95 and .87, respectively, in the Dutch study28 and .94 and .81, respectively, in this study. Validity and reliability data of the shortened anxiety and depression scales are not available so far.
Data Management and Statistical Analyses
The study group included parents with a maximum of 25% of the perceived-support, anxiety, and depression scales unanswered. In the perceived-support scale, such missing values at T1 were as follows: 1 parent had left item 7 unanswered, 1 had left item 8, and 1 had left items 7 and 8 (items presented in Table 1). At T2, 1 parent had left item 1 unanswered. Imputations for these missing values were done regarding analyses including the entire scale scores for perceived support, anxiety, and depression. These imputations were done by replacing missing values with the individual mean score of the scale. On the other hand, in the analyses of single questions of the perceived-support scale, the use of imputation was not applicable. Consequently, these analyses included only respondents who had answered the item in question. The strategy for handling missing data (exclusion of parents with >25 missing subscale items, and the imputation strategy) followed the lines of common practice.
To avoid the potential bias of dependent observations caused by data emanating from mothers and fathers of the same family, the analysis of possible gender differences included only the 20 pairs and treated them as linked subjects. The SPSS 11.0 software was used for statistical analyses.
CHANGE IN PERCEIVED SUPPORT
The relation between perceived support at the 2 points in time was examined in 3 steps:
1. Paired-samples t test was used to examine any linear change in perceived support, by comparing the individual means for perceived support at T1 and at T2.
2. An individual measure of change in perceived support was computed by subtracting the value of support at T1 from that at T2. This value is referred to as support-difference. A negative value indicates deterioration of perceived support, whereas a positive value implies improvement.
3. Based on the support-difference value, 3 groups were formed: change to the better (a higher score at T2), no change, and change to the worse (a lower score at T2). In the no-change group, a deviation of 1 step on the Likert scale in 1 item was allowed. The label support-change groups was used for this categorization.
PREDICTORS OF CHANGE IN PERCEIVED SUPPORT
Anxiety and depression. The predictive power of emotional distress was investigated in 2 different ways. For the continuous support-difference outcome variable, regression analysis was used, whereas the 3 support-change groups were examined using Kruskal-Wallis test because of small subgroup sizes.
Gender. To avoid the potential bias of dependent observations caused by data emanating from mothers and fathers of the same family, a test for related samples was used for analyzing the impact of parental gender on support-difference and the support-change groups. These analyses included mothers and fathers from the 20 couples where both parents had participated in the study and were done using Wilcoxon signed rank test.
SINGLE QUESTIONS OF THE PERCEIVED-SUPPORT SCALE
In addition, we wanted to analyze any change in the various aspects of support represented by the single questions of the perceived-support scale. Such explorative examination was done after dichotomizing item responses. Thus, a response of "I strongly agree" or "I agree" reflected poor support, whereas "I disagree" or "I strongly disagree" indicated good support. A cross-tabulation of the dichotomized answers at T1 and T2 yielded 4 "profiles" regarding each question, as follows:
Poor support, no change: responses indicating poor support at both T1 and T2.
Poor support, change to the better: response indicating poor support at T1 and good support at T2.
Good support, change to the worse: response indicating good support at T1 and poor support at T2.
Good support, no change: responses indicating good support at both T1 and T2.
Results
Scores of perceived support at T1 and T2 and of anxiety and depression (both only at T1) are presented in Table 2. Minimum and maximum values for perceived support were 1.63 and 4.00, respectively, at T1 and 1.50 and 4.00, respectively, at T2. The minimum and maximum values for anxiety were 1.71 and 3.82, respectively, and for depression, 1.00 and 3.00, respectively. The skewness values indicate that all 4 measures were normally distributed: perceived support at T1, -0.47; perceived support at T2, -0.45; anxiety, -0.42; and depression, -0.71.
Change in Perceived Support
The group mean for perceived support at follow-up (T2) was significantly lower than the mean for perceived support at the initial assessment (T1), as indicated by the paired-samples t test (t = 3.03; 95% confidence interval = 0.075-0.368; n = 51; P = .004). Specifically, on a group level, parents reported poorer perceived support at the second assessment.
Support-difference (the individual difference in perceived support between T1 and T2) ranged from -1.75 to 0.50, with a negative value denoting a change toward a poorer perceived support. A graphic representation of support-difference, which may further elucidate this outcome, is presented in Figure 1. The figure shows the distribution of markers (representing individuals) indicating the variation in the relationship between support reported at T1 and at T2. Parents reporting equal levels of support at T1 and T2 are grouped along the diagonal (line) from the lower left to the upper right corner. In cases where parents reported a better perceived support at T2 compared with T1, a marker is shown above the diagonal, whereas parents who reported a poorer perceived support at T2 compared with T1 are represented by markers below the diagonal.
Subsequently, the 3 support-change groups were examined. The groups included change to the better (a higher score at T2), no change, and change to the worse (a lower score at T2). Examination of the support-change groups revealed that a better support at T2 was reported by 25% (n = 13) of the parents. For 27% of the parents (n = 14), support was perceived equally at T1 and T2, whereas 47% (n = 24) stated a perception of poorer support at T2. That is, almost twice as many parents reported a decline in perceived support compared with those whose perceived that support improved. However, chi-square tests demonstrated that the difference in size between the groups reporting improvement and deterioration, respectively, did not reach statistical significance ([chi]2 = 3.27; P = .071).
Predictors of Change in Perceived Support
EMOTIONAL DISTRESS
Depressive symptoms were found to be associated with support-difference: more depressive symptoms at T1 predicted a greater decrease in perceived support over time (n = 51; F = 4.93; P = .031). In contrast, level of anxiety at T1 could not predict magnitude of support-difference (n = 51; F = 1.26; P = .267).
The 3 support-change groups did not differ regarding level of emotional distress at T1 (anxiety [chi]2 = 0.28, P = .867; depression [chi]2 = 2.81, P = .246). Nevertheless, a pattern could be distinguished, as anxiety and depression at T1 were lowest in the group reporting improved perceived support (mean level of depression = 2.17; anxiety = 2.70), intermediate in the group reporting no change (depression = 2.20; anxiety = 2.81), and highest in the group showing a change to poorer perceived support (depression = 2.36; anxiety = 2.84).
PARENTAL GENDER
According to the pairwise analyses of the 20 complete couples in the study group, gender influenced neither the distribution in the 3 support-change groups (P = .763) nor the support-difference value (P = .952).
In the entire group, 9 (31%) mothers were assigned to the better-support group, 8 (28%) to the no-change group, and 12 (41%) to the poorer-support group. The corresponding values for fathers were the following: 4 (18%) experienced better support, 6 (27%) experienced no change, and 12 (55%) experienced poorer support. Mean value of support-difference was -0.21 for mothers and -0.24 for fathers.
Single Items of the Perceived-Support Scale
The analyses of the separate questions of the scale (responses dichotomized) indicated that a majority of the parents perceived good support at T1 and T2 regarding most items (Table 1).
Moreover, these analyses confirmed that decline in perceived support seems to be more common than improvement. According to chi-square tests, this finding was statistically significant for question 2, indicating an increased feeling over time of one's extraordinary needs being socially unmet ([chi]2 = 5.00, P = .025); question 3, indicating an increased feeling of isolation of the family because of the child's cancer ([chi]2 = 5.00, P = .025); and question 6, indicating increased experience of social disappointment because others do not understand your situation ([chi]2 = 4.50, P = .034). For question 1, a tendency in the same direction was found, pointing to an increased feeling of difficulty to find real friends (P = .052). Regarding the remaining questions, no change was indicated: question 4 ("Sometimes I feel very lonely"), [chi]2 = 0.69, P = .405; question 5 ("Sometimes I feel that one is alone in the world after all"), [chi]2 = 0.08, P = .782; question 7 ("People just do not understand that there are still a number of worries left after treatment has ended"), [chi]2 = 0.53, P = .467; and question 8 ("Sometimes it is very difficult to make long-lasting contacts"), [chi]2 = 0.22, P = .637. A look at Table 1 indicates an explanation of the absence of a statistically significant change in perceived social support signaled by these latter items: feelings of loneliness and feelings that others do not fully understand the conditions caused by the child's cancer seem to have been common already during the treatment.
Discussion
In this study, we longitudinally investigated subjectively perceived social support among parents of children with cancer. Parents were initially examined within the first half year of the child's cancer treatment, and a follow-up assessment was done 1 to 2 years later. Cancer treatment was completed at follow-up. Results indicate that some change in perceived support had occurred in a majority of the parents, and this change was most often a decline. One fourth of the parents reported increased satisfaction regarding perceived support. At the same time, almost half of the group reported a poorer perceived support at follow-up. Furthermore, a deterioration of perceived support seemed often to follow after initial emotional distress and especially after depression.
It can be argued that although the individual perception of support is suggested to be rather stable, some normal fluctuations may occur over time. If that were the case, it would be incorrect to assume that modest individual changes in 2 opposite directions should indicate true change. However, the broader distribution of perceived-support scores at T2 indicates that at the second time of assessment, parents indeed more often report good and poor perceived support. This can also be compared with the preliminary findings of a larger variation of scores among parents of children who had survived cancer than among reference parents of children who never had cancer.23 Together, these findings endorse the assumption that the childhood cancer experience may initiate a change in parental perceived support along either of 2 trajectories (one involving growth of faith in support and the other involving a decrease).
Perception of support availability can be supposed to represent a part of a person's existential outlook and general cognitive belief systems. Earlier research on the issue is sparse, but clinical experience suggests that a change of existential outlook and faith in life may take place when parents have experiences of cancer in a child. The cancer experience may entail personal growth and positive change. Parents often say that they have come to appreciate the small things in life.30 Several spontaneous comments of parents in our studies also reveal such general experiences of personal growth, for example, "I can never be 'my old self'! I grow and change 'thanks to' the disease" and "I want to remain this 'wise' me, which I have become from this experience." However, although many parents confirm positive growth and a greater appreciation of life, this is certainly not true for all. This is exemplified by one parent explaining, "Sometimes I think that I still live in a state of shock![horizontal ellipsis]I feel now that life is so unfair." Chesler and Parry31 refer to a father, saying, even more bitterly, "This business about 'coming out of it stronger' is a bit of rubbish." Similarly, parents' perceptions of loneliness and social support seem to be affected. One father in our study said, "The contact with relatives and friends has become better." On the other hand, Grootenhuis and Last32 performed a single-assessment cross-sectional study and noted higher levels of loneliness among mothers more temporally distant from their child's diagnosis. The authors propose that mothers of children with cancer have the feeling that other people do not understand what they have gone through. This assumption is reinforced by parents' spontaneous comments in our study. For instance, one father said, "During the period of illness it became obvious who are your real friends," and one mother said, "There are a few friends who have been there all the time. I was very disappointed when I found out that the ones I thought would be there disappeared. But when it comes down to it, you're all alone." Noteworthy is the finding of Wijnberg-Williams et al17 of no change in dissatisfaction with support over a 5-year period after a child's cancer diagnosis. A tentative interpretation of their finding is that the concept of dissatisfaction with support represents a more stable aspect of personality compared with the concept of perceived social support.
The complex associations between change in perceived social support and reactions to psychological trauma has indeed been explored in other areas. However, to our knowledge, it has not previously been described in the context of parental reactions to the experience of their child's cancer. According to our findings, emotional distress of parents was associated with change in perceived support. Initial depressive symptoms predicted decline in perceived support, and a tendency in the same direction was also found for anxiety. In summary, relying on our and previous findings, we may propose that, in the long run, depression can negatively influence parental perceived support. Furthermore, this influence may be related to the cognitive aspects of depressive mood, which affect appraisal of social support,21,22 as well as social behaviors.19,20 It has also been proposed that perceived support may deteriorate directly through the experience of a traumatic event and indirectly through the loss of available support resources in the face of traumatic or chronic stress.14-16 These speculations have some theoretical support and can be seen as partly supported by our findings. Our study, however, was not designed to-and was underpowered to-test these specific hypotheses.
The analyses of single items of the perceived-support scale give us a more detailed picture of perceived support among parents of children with cancer. The responses to the items "Sometimes, it is very difficult to make long-lasting contacts" and "People just do not understand that there are still a number of worries left after treatment has ended" indicate quite a substantial change both for the better and for the worse. However, the many affirmative answers to the latter context-related statement indicate that this is a problem for a substantial proportion of the parents during treatment and after the end of treatment. Together with the 2 items mentioned above, the item "It is hard to find real friends" represents the greatest change in negative direction. The following statements are also remarkable: "When you really need someone, there is no one there" and "Because of the disease of my child, we as a family have become left on our own." Regarding these items, all or almost all parents initially reported good support. However, at follow-up assessment, 10% of the parents showed a negative development. For the item "Because nobody understands, it is better to keep your worries about your child to yourself," a similar profile was found. In conclusion, findings indicate that perceived support deteriorated more often than improved from the beginning of treatment to off-treatment. Furthermore, this was true regarding issues related to the child's disease and regarding the more general existential issues.
Study Limitations
Despite the interesting findings, a number of limitations also need to be mentioned. One is the relatively small sample, which prevents us from reliably drawing generalized conclusions based on the findings. The sample size also reduces the power of statistical comparisons of subgroups. In addition, although the number of analyses in this study was modest, one should bear in mind the possibility of type I errors. Furthermore, reliance on questionnaires with preformed definitions of problem areas and forced response formats entails the possibility of overlooking important additional aspects of the concept in focus. Finally, a measure of the perceived support before the child was diagnosed with cancer was impossible to obtain, and therefore, we do not know the course of any possible change in perceived support from that point on.
Implications for Practice
The clinical implications of the findings extend to aspects that concern nursing and care. Firstly, the pediatric oncology nurse, as well as other professionals engaged in the day-to-day care of families, has a vital role in encouraging parents' confidence in support and, frequently, in providing emotional support and advice. This may be very valuable because received support serves as a protection of perceived support.14 Secondly, the daily contact with the families enables nurses to identify early on parents in need of supplementary supportive interventions to deal with depression and a tottering faith in life. To this end, nurses should benefit from appropriate guidance and training to distinguish parents at particular risk. Furthermore, for people coping with more severe emotional distress, interventions aiming at merely providing social support have generally not been sufficient for a lasting improvement of mental health.33 In these cases, interventions aimed at modifying interpersonal functioning and improving the perception and appreciation of support from others may, on the other hand, result in long-term positive effects.33
Implications for Research
Suggestions for further research include studying the stability of deterioration and improvement in perceived support. Furthermore, the inclusion of a larger study group would permit the analysis of other factors, possibly relevant for change in perceived support, such as disease-related and socioeconomic factors. Additional targets for forthcoming studies should be the identification of care approaches efficient for strengthening parents' perceived support.
Conclusions
The qualitative analyses and the general direction of the results indicate that perceived support often diminishes over time in parents of children with cancer. Explicitly, the confrontation with cancer of one's own child seems to be a potential threat, which, for many parents, instigates a decrease in perceived support from active treatment to off-treatment. Early depressive symptoms seemed to predict a decrease in perceived support. At the same time, perceived support is an important buffer against distress.4,5,8,34 This complex association between perceived support and distress indicates that maintaining perceived support should be a central goal of nursing and psychosocial care of parents of children with cancer.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the time and energy contributed by the participating parents.
References