Parents of Children With Cancer: At-Risk or Resilient?

Abstract

Objective To examine adjustment in parents of children with cancer using a design that minimizes focusing effects and allows for direct comparison with parents of healthy children. Method Parents of 305 children with cancer and a demographically similar sample of 231 parents of healthy children were evaluated using diagnostic interviews for posttraumatic stress disorder (PTSD), and questionnaire measures of posttraumatic stress symptoms (PTSS) and psychological growth (PG), as well as measures of global psychological functioning. Results Rates of current and lifetime PTSD in parents of children with cancer were low, and did not differ from comparison parents. Likewise, levels of PTSS were not significantly different from comparison parents, but differed as a function of time since diagnosis, with parents of children who were ≥5 years from diagnosis reporting significantly lower PTSS than comparison parents. PG was higher in parents of children with cancer than in comparison parents regardless of time since diagnosis. Conclusion Parents of children with cancer demonstrate resilience to this challenge.

Having a child with cancer, and all that it entails, from diagnosis, through treatment, and hopefully to survivorship, has long been considered a traumatic event and one of the most severe stressors a parent can experience (Kazak, 1998; Stuber, Kazak, Meeske, & Barakat, 1998). The adjustment of parents has been a focus of research, concerned both with parental functioning as a primary outcome, and for the potential influence of parental functioning on the adjustment of the child with cancer (Ljungman et al., 2014; Vrijmoet-Wiersma et al., 2008). The most common parental outcome examined over the past 20 years has been posttraumatic stress disorder and related symptoms (PTSD/PTSS, Bruce, 2006; Ljungman et al., 2014). Early studies reported a prevalence of current PTSD from 6.2% to 25%, and lifetime prevalence of 27% to 54% in parents (predominantly mothers) of children with cancer, a marked 5- to 10-fold increase relative to general population surveys (Bruce, 2006). Subsequent studies have focused on PTSS, and have generally reported substantial symptomatology, but these studies have lacked control or comparison groups of parents of healthy children, with only a few exceptions (Barakat et al., 1997; Brown, Madan-Swain, & Lambert, 2003). A more recent study using an acquaintance control design failed to demonstrate any evidence of increased PTSS in parents of children with cancer relative to parents of healthy children (Jurbergs, Long, Ticona, & Phipps, 2009).

To the extent that having a child diagnosed with cancer represents a potential trauma, it also represents an opportunity to observe parents who demonstrate exceptional strengths in coping and adjustment, and thus resilience to such major life stress. However, most research has been focused on maladjustment and psychopathology, and understandably, on identification of those parents at highest risk, so that they can be targeted for intervention services (Kazak et al, 2012). But the search for pathology, however well intentioned, may unwittingly lead to bias in research design, which can inflate estimates of maladjustment, such as PTSS, and overlook the resilience of many parents in these difficult circumstances. In describing the literature on child adjustment to cancer, Phipps et al. (2014) pointed to “focusing effects” and lack of appropriate control comparisons as factors that have slanted the literature towards pathologic outcomes. The same can be said for much of the literature examining parental adjustment. Recent reviews have demonstrated that studies of parental response to childhood cancer have rarely included comparisons with parents of comparably aged healthy children, and in regards to the outcome of PTSS, essentially all prior studies involve investigator specification of the child’s cancer as the event meeting DSM-IV A1-criteria before administration of survey questionnaires (American Psychiatric Association [APA], 1994; Bruce, 2006; Ljungman et al 2014).

Orienting parents to their child’s cancer as the identified trauma in assessing PTSS has been an understandable approach, since researchers are concerned with cancer-related symptomatology, but it also creates a potential focusing effect (Kahneman, Krueger, Schkade, Schwarz, & Stone, 2006), priming the parent to think of their experience as traumatic, and thereby potentially biasing their subsequent responses. Although it may not be possible to completely eliminate focusing effects in research design, minimizing such cues is desirable, since the impact on subjective symptom reporting can be substantial (Kahneman et al., 2006; Smith, Schwarz, Roberts, & Ubel, 2006; Strack, Martin, & Schwarz, 1988). In addition to creating focusing effects, designs that specify the child’s cancer as the precipitating traumatic event also make it difficult to identify an appropriate comparison group. Perhaps as a consequence, few studies include them. However, given that some level of distress is to be expected in any community sample, one cannot assume that all symptoms reported result from the cancer experience. Moreover, as the demands of parenting can be quite stressful even in the best of circumstances, comparing the psychological functioning of parents of children with cancer to a similar group of parents of healthy children is essential to our understanding of the impact of childhood cancer on parental adjustment.

The present study examined psychological adjustment in a cross-sectional sample of parents of children with cancer, using a design that minimized focusing effects by avoiding a specific cancer orientation, or suggestion that their cancer experience has been traumatic. This allows for identical assessment procedures with a comparison sample of parents unaffected by childhood cancer. With this design, we aim for a more unbiased estimate of parental response, and an appreciation for the full range of parental outcomes from distress and pathology to resilience.

Resilience has been conceptualized in a number of ways, and may be viewed as a trait, a process, or an outcome (Southwick, Bonnano, Masten, Panter-Brick, & Yehuda, 2014). Focusing on outcomes, some suggest that resilience involves only maintaining or returning to premorbid levels of adjustment following adversity, while others suggest it should include bouncing back with improvement and measureable positive outcomes (Southwick et al., 2014). We prefer a conceptualization of resilience similar to Bonnano, as a healthy or adaptive psychological outcome (i.e., minimal distress and disruption in function) over the passage of time following exposure to adversity or potentially traumatic events (Bonnano, 2004; Bonnano, Westphal, & Mancini, 2011; Southwick et al., 2014). The preponderance of research suggests that resilience is the modal outcome across numerous and diverse potentially traumatic events (Bonnano et al., 2011; Deshileds, Tibbs, Fan, & Taylor, 2006; LeBrocque, Hendrikz, & Kenardy, 2010; Masten, 2011, 2014; Moskowitz, Folkman & Acree, 2003; Norris, Tracy & Galea, 2009; Southwick et al., 2014) and we see no reason to suspect this would not also be true among parents of children with cancer.

Beyond resilience, there is some evidence to suggest that adversity may also promote benefits or psychological growth (PG), and this may be particularly true for the challenge of serious illness (Joseph & Linley, 2006; Park & Helgeson, 2006; Tedeschi & Calhoun, 2004). This has been described using various labels, including benefit-finding and posttraumatic growth, but all refer to positive life changes following adversity. Here we refer simply to PG. PG has been more widely studied in adults with cancer, but has been demonstrated in parents of children with cancer, and there is evidence that it occurs with greater frequency in these parents than in parents of children with other serious illnesses, such as diabetes (Barakat, Alderfer, & Kazak, 2006; Hungerbuehler, Vollrath, & Landolt, 2011; Jim & Jacobsen, 2008; Lindwall et al., 2014). PG and resilience are theoretically distinct, but related constructs, and findings regarding their empiric relationship have been inconsistent, in particular, those examining correlations of PG and PTSS (Hungerbuehler et al., 2011; Klosky et al., 2014). Assessment of resilience involves administration of general measures of distress or pathological outcomes, such as depression, anxiety, or PTSS, with demonstration of low or minimal symptom levels, whereas assessment of PG uses measures designed specifically to tap perceived positive changes such as finding new personal strength or increased family closeness (Antoni et al., 2001; Tedeschi & Calhoun, 1996). Both PG and PTSS are included here to provide a more comprehensive set of measures to assess adjustment outcomes of parents who must manage the stress of childhood cancer.

The objectives of this study were to examine parental adjustment in the context of childhood cancer while addressing previous research limitations. We included a demographically similar comparison group, used identical assessment procedures for both groups, including diagnostic interviews for current and lifetime incidence of PTSD, minimized focusing effects by avoiding a cancer orientation while still allowing for a cancer-specific assessment, and included measures of PG to obtain a more comprehensive assessment of the range of parental outcomes, both positive and negative. We hypothesized that (1) rates of PTSD would be no higher in parents of children with cancer than in comparison parents; (2) the resilience of parents of children with cancer would be demonstrated by low overall levels of PTSS, and adjustment scores that do not differ from comparison parents; and (3) based on the findings of Hungerbuehler et al. (2011), and extrapolating from studies of the cancer patient, we predicted that measures of PG would be higher in parents of children with cancer than comparison parents, and further, that this difference would be greater when referencing cancer-specific events.

Method

Participants

Cancer Group

Parents of children who were treated for malignancy at a major pediatric cancer center were recruited during the child’s routine outpatient visits. Eligibility was based on the child patient characteristics and included: (1) patient age 3–17 years; (2) diagnosis of any malignancy; (3) ≥1 month from diagnosis; (4) English speaking; and (5) child with no significant cognitive or sensory deficits. For each child recruited, we sought the participation of one parent, which could be mother, father, or, in a few instances, another custodial relative. Participants were recruited in four strata based on the child’s elapsed time from diagnosis (1–6 months; >6 months to 1.99 years; 2–4.99 years; ≥5 years). Of 444 approached, 309 (70%) agreed to participate. Participants and nonparticipants did not differ by child age, gender, race/ethnicity, time from diagnosis, or cancer diagnosis. Four withdrew before providing data, leaving a final sample of 305.

Comparison Group

Comparison parent eligibility included: (1) parent of child age 3–17 years; (2) no history of serious illness in reference child; (3) English speaking; and (4) no cognitive or sensory deficits in the reference child. Recruitment was conducted primarily through public schools in a three-state area surrounding the hospital. With approval of school system superintendents and school principals, permission slips were distributed, requesting consent to contact the family if the student was a demographic match for a patient who had been enrolled on study. The returned slips contained contact information along with the child’s age, gender, race/ethnicity, and parental education and occupation. This information was used to create a pool of potential comparison participants, who were subsequently contacted. Potential comparison participants who provided an appropriate demographic match were contacted initially via telephone by study research assistants. For those parents who expressed interest in participating, an appointment was made for them to be seen in the Psychology Department of the hospital, where consent/assent were obtained and study procedures conducted in the same manner as for cancer group participants. Parents of younger children were recruited in a similar manner through day care centers and preschool programs in the area. Matching was based on child, rather than parent, characteristics. Participants were matched on child age (within 1 year), gender, race/ethnicity, and as close as possible on family socioeconomic status. Of 265 families contacted, 231 (87%) agreed to participate.

The demographic and medical background of the sample is summarized in Table I. Groups did not differ in percentage of mothers and fathers participating. However, there were more in the “other” category in the cancer group χ² (2, N = 536) = 10.0, p < .01). Groups did not differ in terms of child age, gender, or race/ethnicity. Using the Barratt Simplified Measure of Socioeconomic Status (SES, Barratt, 2006), there was a significant between-group difference, with fewer parents in the cancer group from the higher SES strata, and more in the lower SES strata relative to comparison parents χ² (2, N = 535) = 11.5, p < .01). The diagnostic breakdown of the patient sample is generally representative of the population served by the hospital. There were nearly equal numbers of children who were on therapy or had completed therapy, and there was a broad distribution across the time since diagnosis strata. We included a number of patients (13.8%) who had experienced a relapse. Patients were also categorized using the Intensity of Treatment Rating Scale 2.0 (ITR-2, Werba et al., 2007), a 4-point ordinal scale. The majority of patients were categorized as having received moderate or very intensive therapies (Table I).

Table I.

Demographic and Medical Variables

	Cancer		Control
	N	%	N	%
Parent participant
Mother	252	82.6	206	89.2
Father	38	12.5	24	10.4
Other	15	4.9	1	0.5
SES strata
I & II	81	26.6	92	39.8
III	93	30.6	66	28.6
IV & V	130	42.8	73	31.6
Race
White	221	72.5	177	76.6
Black	67	21.9	46	19.9
Other	17	5.6	8	3.5
Child gender
Male	159	52.1	118	51.1
Diagnosis
ALL	76	24.9
Other leukemia	20	6.6
HD/NHL	37	12.1
Solid tumor	119	39.0
Brain tumor	53	17.4
Time since diagnosis
<6 months	72	23.6
6 months–2 years	78	25.6
2–5 years	92	30.2
>5 years	63	20.7
On/Off therapy
On	147	48.2
Off	158	51.8
Relapse status
No	263	86.2
Yes	42	13.8
ITR-2
Least intensive	21	7.3
Moderately intensive	105	36.3
Very intensive	90	31.1
Most intensive	73	25.3

Procedure

The described research was part of a larger study examining child, parent, and family adjustment to childhood cancer. As part of our attempt to minimize focusing effects, the consent form indicated the study was on “stress, coping and adjustment in children and their parents,” and did not specify a focus on cancer or trauma. After informed consent was obtained, parents completed a battery of questionnaires. Participants spontaneously identified their most stressful or traumatic event without any orientation to cancer. Diagnostic interviews for PTSD were based on the parents’ self-identified event. In order to meet full criteria for PTSD, parents must have identified an event that met DSM-IV A criteria, as well as symptom criteria B-D and functional criteria E and F. However, PTSS were assessed from all parents based on their identified event, regardless of whether it appeared to meet A-criteria. Questionnaire measures of PTSS and PG were obtained in reference to the parents’ spontaneously identified most stressful event. However, for parents in the cancer group who did not identify an event related to their child’s cancer, when all other study procedures were completed, they were oriented to their child’s cancer and asked to identify the aspect of this experience that they found the most stressful. They were then asked to complete the PTSS and PG measures again in relation to this cancer experience. This allowed for the initial procedures to be identical across groups, so that they could be equivalently compared on the basis of what they spontaneously identified as their most traumatic event, but also allowed for a cancer-specific assessment without a preemptive focusing on cancer, or suggestion that the cancer experience was traumatic.

Measures

Clinician Administered PTSD Scale

The Clinician Administered PTSD Scale (CAPS) is a structured diagnostic interview for PTSD (Weathers, Keane, & Davidson, 2001; Weathers et al., 2004). It evaluates current and lifetime criteria for PTSD, with ratings of the frequency and intensity of symptoms as outlined in the DSM-IV. Assessment begins with a checklist of life events that meet DSM-IV A-1 criteria. Interviews proceed based on the event the respondent identifies as their most traumatic, if it meets A-criteria. All parents in the cancer group were considered to have met A-1 criteria, even if they identified an event other than their child’s cancer as their most traumatic. The CAPS is one of the most widely used PTSD interviews and has been shown to have excellent psychometric properties in a variety of populations. In the current study, using six raters, average absolute agreement across nine inter-rater dyads was 97% for current PTSD (average Kappa = 0.87), and for lifetime PTSD, agreement was 100%.

Impact of Events Scale, Revised

The 22-item Impact of Events Scale, Revised (IES-R) measures PTSS in three subscales, Intrusion, Avoidance, and Hyperarousal, corresponding to DSM-IV clusters B, C, & D. It is one of the most widely used instruments in the study of PTSS, and reliability and validity are well documented. Parents completed this measure as a self-report of their own PTSS in reference to the event they spontaneously described as their most traumatic (Weiss & Marmar, 1997). As mentioned above, for parents in the cancer group who identified a noncancer event initially, following the completion of all other study procedures, they were oriented to their child’s cancer and asked to complete this, and the PG measures a second time. For this study, only the total IES score was used, and internal reliability (Cronbach’s α) was .94.

Brief Symptom Inventory

The Brief Symptom Inventory (BSI) is a 53-item measure based on the SCL-90-R that was used to assess parental psychological symptoms (Derogatis & Melisaratos, 1983). T-scores can be obtained on nine different symptom clusters and three global subscales. Here we report only on the Global Severity Index (GSI), using the adult nonpatient normative sample, to obtain gender normed T-scores. Excellent reliability data have been reported (test–retest reliability for GSI of .90), and convergent and divergent validity (Derogatis & Melisaratos, 1983).

Positive and Negative Affect Schedule

The Positive and Negative Affect Schedule (PANAS) is composed of two 10-item scales, one measuring positive affect, one measuring negative affect (Watson, Clark & Tellegen, 1988). Here, it was used in the trait format, asking parents to respond based on how they “generally feel.” The instrument has excellent reliability and demonstrated utility with a variety of adult populations (Watson, Clark, & Tellegen, 1988). In the current study, internal reliability (Cronbach’s α) was .88.

Benefit Finding Scale

This 17-item scale was initially developed to assess perceived benefits of women with breast cancer (Antoni et al., 2001). Items referred to positive changes as a result of illness, such as “helped to make me a stronger person,” “helped me make some new friends,” etc. Each item is responded to on a 5-point scale from “Not at all” to “Very much.” It has also been used to assess perceived benefits of caregivers of cancer patients (Kim, Schulz, & Carver, 2007), and in a study with parents of children undergoing stem cell transplantation (Lindwall et al., 2014). Here, parents responded to items based on the same event they identified in completing the IES-R. Cronbach’s α in the current study was .92.

Posttraumatic Growth Inventory

This 21-item scale is the most widely used measure of growth/benefit following trauma in adults. Items describing change resulting from the identified event (e.g., “have a greater appreciation for the value of my own life”; “know better that I can handle difficulties”; “have a greater sense of closeness with others”) were rated on a 6-point scale ranging from 0 (did not experience this change) to 6 (experienced this change to a great degree) (Tedeschi & Calhoun, 1996). It contains five subscales: relating to others, new possibilities, personal strength, spiritual change, and appreciation of life. However, for the present study, only the total score was used (Cronbach’s α = .95) and again, parents responded in reference to the same event identified on the IES-R and Benefit Finding Scale (BFS).

Life Events Scale

The Life Events Scale (LES; Johnston et al., 2003) used here was a 28-item version adapted from the scale described by Johnston, Herrera, Steele and Phipps (2003), which in turn was adapted from the original LES of Coddington (1972). It is composed of significant events, some of which would meet A-1 criteria of the DSM-IV and others which may not, but would be expected to have a significant stressful impact (e.g., divorce, natural death of family member). Parents indicated whether they had experienced each event, and approximately when it occurred. Items are not weighted and a simple sum was obtained as a measure of cumulative life events.

Analytic Plan

The first outcome to be examined was the percentage of parents who spontaneously identified their child’s cancer as their most stressful event. We next compared the rates of parents meeting full criteria for PTSD (both current and lifetime) in cancer and comparison groups. Our sample provided 80% power to detect a between-group difference of approximately 3% based on Fisher’s exact test, assuming very low rates (i.e., approaching 0%). We then examined between-group differences on PTSS and PG measures. Cancer group outcomes were characterized in three ways: (1) PTSS and PG scores in response to the spontaneously reported most traumatic event, regardless of whether it was related to the child’s cancer; (2) PTSS and PG scores related to the child’s cancer; and (3) PTSS and PG scores reported in response to noncancer events. Each of these was compared with the comparison group means. Group comparisons were conducted with linear regression models controlling for SES and life events, owing to observed group differences on these factors. For full group comparisons, the current sample provided 80% power to detect an effect of 0.24 SD, or slightly more than 4 points on the IES-R measure, 3 points on the BFS, and 5.5 points on the Posttraumatic Growth Inventory (PTGI). Finally, we examined the impact of time since diagnosis and on/off therapy status on PTSS and PG outcomes. This was conducted with linear regression controlling for life events and SES. The comparison group was used as the reference group for post hoc comparisons.

Results

Identification of Cancer as Traumatic Event

Even with efforts to avoid orientation to cancer, 77.7% of parents in the cancer group identified their child’s cancer as their most traumatic event. The proportion identifying a cancer-related event differed by time since diagnosis χ² (3, N = 305) = 26.6, p < .001), with 58.7% of parents whose child was 5 or more years from diagnosis doing so, in comparison with 82.6% of those whose child was diagnosed earlier. The proportion identifying a cancer event did not differ by child age, gender, race-ethnicity, participating parent, on versus off treatment status, relapse status, or treatment intensity. There were marginal differences based on SES, t(303) = 2.20, p = .029, indicating lower SES scores in those identifying a noncancer event. There was also a marginal difference based on diagnostic category (Fisher’s exact test, p = .026), with parents of children with Hodgkin’s disease and Non-Hodgkin’s lymphoma slightly more likely to identify a cancer-related event, and those with solid tumor slightly less so. Of the 68 parents (22.3%) who identified a noncancer event, 69.1% reported an event that met DSM-IV A-1 criteria. This did not differ from the 69.3% of comparison parents whose identified event also met A-1 criteria.

Rates of Current and Lifetime PTSD

Diagnostic interview on the CAPS identified five parents (1.6%) in the cancer group who met current criteria for full PTSD (Table II ); in four of these, the precipitating event was the child’s cancer. Using logistic regression, controlling for differences in SES, this rate was not significantly different from the comparison group, where two cases (0.9%) of current PTSD were identified. SES was not a significant covariate (p = .9). A total of 22 parents in the cancer group (7.3%) met lifetime criteria for PTSD, which again did not differ significantly from the 9.5% of comparison parents who did so. SES was a significant covariate in this comparison (p = .011). The precipitating events for those who met full PTSD criteria in each group are listed in Table II. Although the majority of parents meeting lifetime criteria in the cancer group identified their child’s cancer as the precipitating event, 41% resulted from a noncancer event. The noncancer events precipitating PTSD were comparable in the two groups.

Table II.

Cases Meeting Full Criteria for PTSD

	Cancer group (N = 305)		Comparison (N = 231)
	N	%	N	%	O.R. (95% CI)	p
Current	5	1.6	2	0.9	1.88 (0.37–9.96)	.457 NS
Lifetime	22	7.3	22	9.5	0.63 (0.33–1.20)	.158 NS
Precipitating event	Current (5)		Current (2)
	Child’s Cancer – 4		Abuse – 1
	Other illness (family) – 1		Military combat (Iraq) – 1
	Lifetime (22)		Lifetime (22)
	Child’s Cancer – 13		Auto Accident – 2
	Divorce (life threat) –2		Divorce (life threat) – 1
	Abuse – 2		Abuse – 3
	Military combat (Viet Nam) – 1		Military combat (Iraq) – 1
	Death in family – 1		Death in family – 8
	Sexual assault/rape – 1		Sexual assault/rape – 3
	Serious illness (family) – 2		Serious illness (family) – 2
			Serious illness (self) – 1
			Witnessed murder – 1

Note. OR = odds ration; CI = confidence interval.

Adjustment and Growth Outcomes

Prior to testing group differences on adjustment and growth measures, we examined cumulative stressful life events in the two groups, as reported on the LES. Parents of children with cancer reported a marginally higher level of life events (M = 7.4, SD = 3.9) than did our comparison parents (M = 6.7, SD = 3.7; t = 2.18, p = 0.03, d = .19). This difference may be largely explained by the fact that having a child with a serious or life-threatening illness is an item on the LES. The LES score was thus included as a covariate (along with SES) in all subsequent comparisons.

Group comparisons on all outcome measures are summarized in Table III . For the PTSS and PG measures, cancer group scores are reported separately for: (1) reference to the most traumatic identified event, regardless of relation to the child’s cancer; (2) reference to the child’s cancer; and (3) reference to a noncancer event. BSI and PANAS measures were obtained only once from all participants, and thus involve a single comparison. All comparisons included SES and LES scores as covariates. On the IES-R, parents of children with cancer did not differ from comparison parents regardless of whether they were referring to a cancer or noncancer-related event. SES was a significant covariate in all comparisons (all p values < .05), indicating PTSS was greater in lower SES groups. Likewise, LES was a significant covariate in all comparisons (all p values < .001), indicating PTSS was greater with increasing cumulative life events. On the BSI-GSI, parents of children with cancer did not differ from comparison parents, and both SES (p = .022) and life events (p < .001) were significant covariates in the consistent direction. Likewise there was no difference on the PANAS measure of positive affect, where cancer and comparison groups had nearly identical scores. Neither SES nor life events scores were significant covariates.

Table III.

Cancer and Comparison Group Differences on Measures of Adjustment and Psychological Growth

Measure	Cancer group		Healthy comparison group
	N	Mean (SD)	N	Mean (SD)	t	p	Cohen’s d
Adjustment measures
PTSS (IES-R)
Most stressful event	292	24.7 (18.0)	226	21.2 (17.9)	1.3	.176	0.20
Cancer event	291	23.1 (17.8)			0.4	.698	0.10
Noncancer event	64	25.6 (20.3)			0.7	.488	0.24
Global adjustment
BSI-GSI	299	54.7 (11.4)	230	53.4 (10.4)	0.3	.755	0.12
Positive affect (PANAS-P)	292	36.3 (6.5)	226	36.7 (5.8)	0.7	.515	−0.06
Growth measures
PG (BFS)
Most stressful event	299	68.1 (12.5)	231	57.7 (14.5)	8.3	<.001	0.77
Cancer event	298	69.1 (11.9)			9.4	<.001	0.87
Noncancer event	66	61.6 (13.8)			1.5	.140	0.28
PG (PTGI)
Most stressful event	299	72.1 (22.3)	228	58.2 (24.8)	6.1	<.001	0.60
Cancer event	294	72.9 (22.2)			6.6	<.001	0.63
Noncancer event	67	69.3 (23.8)			2.8	<.01	0.46

Note. PTSS = posttraumatic stress symptoms; IES-R = Impact of Events Scale, Revised; BSI = Brief Symptom Inventory; GSI = Global Severity Index; PANAS-P = Positive and Negative Affect Schedule, positive affect; PG = psychological growth; BFS = Benefit Finding Scale; PTGI = Posttraumatic Growth Inventory.

In regards to PG outcomes, on the BFS, parents in the cancer group reported significantly greater benefit than comparison parents when referring to their child’s cancer t(525) = 8.3, p < .001, but not when referring to a noncancer event t(293) = 1.5, p = .14, Table III. SES was not a significant covariate in these comparisons, and LES score was a significant covariate only in reference to a noncancer event (p = .027), indicating greater benefit was reported with increasing events. On the PTGI, parents in the cancer group reported significantly greater growth than comparison parents for both their child’s cancer (t(518) = 6.6, p < .001) and noncancer events (t(291) = 2.8, p = .005). SES was not a significant covariate, but LES was in all comparisons (all p values < .05), indicating greater growth with increasing cumulative events. It should be noted that the correlation between the BFS and the PTGI across the full sample was .78 (p < .001), suggesting considerable overlap between these two measures.

Effect of Time Since Diagnosis and Treatment Status

The impact of time since diagnosis on PTSS and PG was examined using the time since diagnosis strata as a categorical variable, and including the healthy comparison parents as the reference group against which each stratum was compared (Table IV ). For these analyses, we used the parental scores referencing the cancer event, as these are the most relevant to such cancer-related predictors, although the results using the most stressful event (i.e., spontaneously identified event) score were nearly identical.¹ On the IES-R, there was a linear pattern of scores declining with increasing time since diagnosis. Parents whose child was diagnosed in the past 6 months reported the highest levels of PTSS descriptively, but still did not differ significantly from comparison parents (t(510) = 1.9, p = .053). In contrast, parents of children who were 5 or more years from diagnosis reported PTSS scores significantly lower than comparison parents t(510) = − 3.0, p = .003. Similarly, parents whose child was still on treatment reported greater PTSS than parents whose child was off treatment, but neither group differed significantly from comparison parents. In contrast, on both the BFS and PTGI, parents in all time from diagnosis groups, as well as parents of children both on and off treatment, reported significantly greater PG than comparison parents. These differences were substantial, with effect sizes approaching .5 or greater on all comparisons (Table 4). We also examined the effect of diagnostic category, relapse status, and treatment intensity group, and none of these were associated with significant differences on PTSS and PG outcomes.

Table IV.

PTSS and PG Outcomes by Time Since Diagnosis and on Versus Off Treatment

Measure	Cancer group		Healthy comparison group
	N	Mean (SD)	N	Mean (SD)	t	p	Cohen’s d
PTSS (IES-R)
Time since diagnosis
1–6 months	68	27.4 (17.9)	226	21.2 (17.9)	1.9	.053	0.34
>6 months–1.99 years	74	24.6 (18.8)			0.9	.360	0.19
2–4.99 years	88	23.9 (16.4)			0.9	.373	0.16
>5 years	61	15.2 (16.2)			−3.0	.003	−0.34
On/Off Treatment
On treatment	137	25.8 (17.1)			1.7	.093	0.26
Off treatment	154	20.7 (18.1)			−0.9	.374	−0.03
PG (BFS)
Time since diagnosis
1–6 months	70	68.0 (14.4)	231	57.7 (14.5)	5.5	<.001	0.71
>6 months–1.99 years	77	71.6 (10.6)			7.7	<.001	1.02
2–4.99 years	91	69.5 (10.9)			7.0	<.001	0.87
>5 years	60	66.5 (11.0)			4.4	<.001	0.64
On/Off treatment
On treatment	144	68.9 (12.9)			8.0	<.001	0.80
Off treatment	154	69.3 (10.9)			8.5	<.001	0.88
PG (PTGI)
Time since diagnosis
1–6 months	70	70.3 (25.9)	228	58.2 (24.8)	3.5	<.001	0.49
>6 months–1.99 years	73	78.5 (19.1)			6.2	<.001	0.87
2–4.99 years	92	72.6 (21.9)			4.8	<.001	0.60
>5 years	59	69.4 (20.6)			3.0	.003	0.47
On/Off treatment
On treatment	142	72.5 (23.7)			5.7	<.001	0.59
Off treatment	154	73.2 (20.7)			6.2	<.001	0.65

Note. The t values, significance, and effect size estimates reflect the comparison of the means of each time since diagnosis, or on/off treatment subgroup in the cancer group with the overall comparison group mean.

Discussion

Assuming that having one’s child diagnosed with cancer is a highly stressful and potentially traumatic event, the current findings point to considerable resilience in parents who face this challenge. Although the great majority of parents identified their child’s cancer as the most traumatic event they had ever experienced, they appear to adjust very well, with little evidence of psychopathology or increased distress. Using a well validated structured diagnostic interview, parents of children with cancer met full criteria for current and lifetime PTSD at rates no higher than comparison parents, and at levels similar to what would be expected in any community sample (Bruce, 2006). Likewise, they show no evidence of elevated levels of PTSS relative to comparison parents of similarly aged children. Moreover, measures of global adjustment on the BSI, as well as positive affect on the PANAS did not differ between groups. These findings are consistent with the framework of resilience as described by Bonnano et al. (2011) and provide further evidence that resilience is the modal response following significant adverse events. Although much of the prior literature has been focused on identifying pathological responses, the current findings suggest that psychopathology is the exception rather than the rule in this parent population.

Consistent with prior research (Jurbergs et al., 2009; Phipps, Long, Hudson, & Rai, 2005), the current findings indicate that PTSS differs as a function of time elapsed since diagnosis and whether the child is on current treatment. In the absence of longitudinal data, the current design provides an approximation which suggests a temporal pattern, with increased PTSS in the period following diagnosis, which decreases gradually over time. Parents whose child was recently diagnosed show higher levels of PTSS, yet still do not differ significantly from comparison parents. In contrast, parents whose child was 5 or more years from diagnosis showed significantly lower levels of PTSS than comparison parents. The levels of PTSS reported by parents of recently diagnosed patients in the current study are comparable with those from prior multisite studies of parents of newly diagnosed patients (Dolgin et al., 2007; Dunn et al., 2012) and of parents whose children were undergoing stem cell transplant (Lindwall et al., 2014). The current study extends those findings, with inclusion of a demographically similar comparison group, and the absence of a significant difference replicates an earlier finding by Jurbergs et al. (2009). The surprisingly low levels of PTSS in parents of long-term survivors also replicates a finding by Jurbergs et al. (2009) and suggests there may be something about the experience of parenting a child with cancer that promotes resilience, or increased stress-resistance over time. A limited literature on resilience in families of children with cancer suggests that parents do experience changes that may strengthen their abilities to cope with stress, including increased social support and ability to rapidly adapt to changing demands (Brody & Simmons, 2007; Grootenhuis & Last, 1997; McCubbin, Balling, Possin, Frierdich, & Byrne, 2002). Perhaps there is a type of stress-inoculation that occurs when parents are forced to struggle with the early challenges of their child’s illness (Meichenbaum, 1993). Others have described this phenomenon, as “toughening” (Dienstbier, 1992), “steeling” (Rutter, 2006), or “thriving” (Carver, 1998). Dienstbier (1992) suggests that stress exposure may have a positive toughening effect when the exposure is limited and manageable, with opportunity for recovery or mastery. Consistent with this view, recent empiric studies have demonstrated a U-shaped relationship between cumulative life adversity and resilience to stress, concluding that in moderation, experiencing adversity contributes to the development of resilience (Seery, Holman & Silver, 2010; Seery, Leo, Lupien, Kondrak, & Almonte, 2013). This suggests that the experience of parenting a child with cancer, while stressful, is a manageable stressor for most parents, which when successfully navigated, may leave them with an increased capacity to handle future stress.

As predicted, parents of children with cancer also reported significantly greater PG than comparison parents. On the BFS measure, parents of children with cancer reported greater benefit only in response to a cancer event and not to a noncancer event; whereas, on the PTGI, they reported greater PG for both cancer and noncancer events. Again, this finding would appear to point to something unique in the experience of parenting a child with cancer that promotes the capacity to find benefit in difficult circumstances. There has been discussion about terminology, and whether benefit finding and growth represent distinct or similar constructs. The correlation between the BFS and PTGI measures suggest considerable overlap, such that they might be considered a unitary construct. Perhaps more importantly, the construct of PG has been criticized, and questions have been raised as to whether self-reported PG reflects genuine change, or just a perception of change that may be illusory (Frazier et al., 2009; Zoellner & Maercker 2006). It may be that PG, rather than indicating actual growth, is more reflective of positive reinterpretation coping (Frazier et al., 2009). Nevertheless, even if the elevated levels of PG are primarily a result of self-serving judgments (Roese & Olson, 2007), they result in a perception of increased well-being, and in combination with the low levels of distress observed, point to the global positive adjustment of this population. Clearly, parents of children with cancer perceive themselves as having low levels of distress, and as having grown or benefitted in some way from their child’s cancer experience.

In addition to demonstrating parental resilience to the stresses of childhood cancer, the current findings might also be seen as raising questions about the magnitude of the experience and whether it is as traumatic as previously thought. Medical advances in the past 30 years have significantly reduced the mortality and morbidity of childhood cancer, and numerous advances in supportive care have reduced the discomforts associated with the illness and its treatment (DeSantis et al., 2014; Pizzo & Poplack, 2010). Some of the early reports of higher levels of PTSD in parents may reflect cohort effects, with the poorer prognosis and greater invasiveness and discomforts of earlier treatment eras. While the diagnosis of a child’s cancer in 2015 undoubtedly remains a stressful and challenging circumstance for parents, the current treatment context, including greater availability of psychological support, may indeed imply a more manageable stressor, with opportunities for parents to develop mastery, rather than pathology.

All assessments of PTSD/PTSS in the current study were based on the DSM-IV, as the study was designed and much of the data collected before the release of the DSM-V (APA, 2013). While there have been some important changes in the recognized symptom clusters for PTSD, perhaps the most significant change relevant to the current study is in regard to the A-criteria. Whereas the DSM-IV explicitly identified diagnosis with a life-threatening illness, or for witnessed events, “learning that one’s child has a life-threatening illness” as meeting A-1 criteria, the DSM-V states that a life-threatening illness is not necessarily considered a traumatic event. Rather, it must involve a sudden and catastrophic event. The language for witnessed events now indicates “a medical catastrophe in one’s child (e.g., a life-threatening hemorrhage)” (pp. 274). This language is subject to interpretation, but raises doubt as to whether it would continue to apply to the diagnosis of cancer in a child, in the absence of a more acute life-threatening event such as an ICU admission. The decline in morbidity/mortality and improvement in supportive care noted above, and the changes in wording of the A-criteria in the DSM-V, both raise questions about the continued application of a trauma model in the setting of childhood cancer. The current findings, which fail to demonstrate any increase in either PTSD or PTSS, would seem to argue that the diagnosis of cancer in one’s child is, by itself, not sufficient as an A-criteria event.

Beyond diagnostic considerations, there are potential clinical implications raised by the current findings. The overall positive adjustment observed, and the low levels of PTSD suggests that systematic application of prophylactic interventions is unnecessary, and would likely be an inefficient utilization of resources, particularly if focused on prevention of PTSD. When providing anticipatory guidance to parents of recently diagnosed children, reassurance can be given that early symptoms decline over time, and that most parents become adept at handling the challenges of their child’s illness. To the extent that parents do gain an increased capacity to handle stress, a promising approach to support for parents of the newly diagnosed is exposure to more experienced parents, which can happen informally, or more formally through something like a parent mentor program. Parents who have had opportunity for some mastery experiences can serve as role models and natural supports for parents initially facing this challenge, with professional intervention reserved for the small subset of parents who demonstrate persistent difficulties.

A significant limitation of the present study is its single-site design, which raises questions about the generalizability of the findings, particularly given their discrepancy with much of the prior literature. However, a number of prior multisite studies that included the present site and examined similar parental outcome measures such as the IES-R, have failed to find any significant site differences (Dolgin et al., 2007; Lindwall et al., 2014; Sahler et al., 2005, 2013). We believe that the current findings are more reflective of the study design, with our attempt to minimize focusing effects and inclusion of a demographically similar comparison group, than to any uniqueness of the site. Our participation rate was moderately low, at 70%. Although participants did not differ from nonparticipants on any of the demographic or medical variables that we could assess, this nevertheless raises the possibility that some of the more distressed parents may have chosen not to participate. Another potential limitation is that our cancer group parents completed measures on the same day as their child had an oncology visit, which may have impacted their responses. However, our approach of collecting all data from both cancer and comparison groups in the hospital provided an identical setting for all assessments. A final limitation is our cross-sectional design. The optimal approach for examining response to a potentially traumatic event is a longitudinal design with repeated assessments following the identified event. Our examination utilizing time since diagnosis groups can provide only an approximation of the effect of change over time. Nevertheless, the observed difference in PTSS between parents of children recently diagnosed versus parents of children 5 or more years from diagnosis, is highly suggestive of a temporal trend, with decline of symptoms over time. Multisite, longitudinal studies examining the adjustment of patients and parents from diagnosis are needed, and should be a priority for future research.

In summary, the current findings document low levels of distress and pathology in parents of children with cancer, and together with the high level of perceived PG observed, provide another example of human resilience and capacity to thrive after adverse events. It appears that there may be elements of the cancer experience that promote subsequent resilience in these parents, through exposure to stressful, but manageable, challenges in the context of a supportive environment. However, much work remains to be done to understand the determinants and mechanisms of resilience in this setting, and delineating pathways to resilience has emerged as a proper focus of future research.

Funding

National Institutes of Health (NIH) R01 CA136782 and the American Lebanese-Syrian Associated Charities (ALSAC).

Conflicts of interest: None declared.