Unrealistic Parental Expectations for Cure in Poor-Prognosis Childhood Cancer

Abstract

BACKGROUND:

Many parents of children with advanced cancer pursue curative goals when cure is no longer possible. To the authors’ knowledge, no pediatric studies to date have prospectively evaluated prognosis communication or influences on decision making in poor-prognosis childhood cancer.

METHODS:

The authors conducted a prospective cohort study at 9 pediatric cancer centers that enrolled 95 parents of children with recurrent or refractory, high-risk neuroblastoma (63% of those who were approached), a condition for which cure rarely is achieved. Parents were surveyed regarding the child’s likelihood of cure; their primary goal of care; the child’s symptoms, suffering, and quality of life; and regret concerning the last treatment decision. Medical records identified care and treatment decisions.

RESULTS:

Only 26% of parents recognized that the chance of cure was <25%. When asked to choose a single most important goal of care, approximately 72% chose cure, 10% chose longer life, and 18% chose quality of life. Parents were more likely to prioritize quality of life when they recognized the child’s poor prognosis (P = .002). Approximately 41% of parents expressed regret about the most recent treatment decision. Parents were more likely to experience regret if the child had received higher intensity medical care (odds ratio [OR], 3.14; 95% CI, 1.31–7.51), experienced suffering with limited benefit from the most recent treatment (OR, 4.78; 95% CI, 1.16–19.72), or experienced suffering from symptoms (OR, 2.91; 95% CI, 1.18–7.16).

CONCLUSIONS:

Parents of children with poor-prognosis cancer frequently make decisions based on unrealistic expectations. New strategies for effective prognosis communication are needed.

INTRODUCTION

Cancer is the leading cause of disease-related mortality in children in the United States.^1,2 For those who die, significant suffering complicates the end-of-life period,^3–6 and intensive treatment remains the norm.³ Greater than one-half of children who die of cancer receive cancer treatment within the last month of life, approximately one-half die in the hospital, and nearly one-half of hospital deaths occur in the intensive care unit (ICU), usually after withdrawal of cardiorespiratory support.³ Those children who receive end-of-life cancer treatment are reported to suffer from more symptoms than those who do not.^4,7 Nevertheless, many parents search for curative therapy in the advanced cancer setting,⁸ perhaps in part because parents often are not well informed regarding prognosis,^9–11 a situation that leads parents to pursue intensive therapy at the end of life,¹⁰ even if it has little chance of efficacy.

Taken together, these issues raise the question of whether improved parental decision making can improve the quality of life children experience near life’s end. Previous work has been challenged by retrospective study design and the heterogeneity of the diagnoses studied. We prospectively enrolled a cohort of parents of children with a single, poor-prognosis cancer, recurrent or refractory high-risk neuroblastoma, using neuroblastoma as a model that might offer insight into the care of all children with life-threatening conditions. Greater than one-half of children with newly diagnosed high-risk neuroblastoma either will never achieve disease remission or will experience disease recurrence after treatment,¹ and the vast majority of these children will die. The 5-year overall survival rate is estimated at 4% after disease recurrence,^12–14 and is only slightly higher in patients with refractory disease.^13,14 Despite poor prognoses, children with recurrent or refractory neuroblastoma have numerous options for care, including chemotherapy regimens of varying intensity, clinical trials, immunotherapy, radionuclide therapy, radiotherapy, and non–cancer-directed treatment focused on symptom management. Therefore, recurrent/refractory high-risk neuroblastoma presents a model for parental decision making, within the setting of a complex array of choices, for children with advanced cancer.

The goal of the current study was to identify communication and care experiences that drive parents’ care goals and decisions. We developed a theoretical model (Fig. 1) in which understanding of prognosis influences the care goals of parents, which in turn influence treatment decisions and outcomes, including the child’s quality of life. We hypothesized that parents’ goals also are influenced by prior experiences such as suffering from symptoms. Finally, we hypothesized that parents continuously reflect on decisions, and that decisions resulting in high treatment burdens may increase regret.

Figure 1.

Theoretical model for the current study.

MATERIALS AND METHODS

We surveyed the parents of children with recurrent/refractory high-risk neuroblastoma who were receiving care at 9 sites (Table 1) from September 2013 to July 2018. Parents were eligible for participation any time after the child’s diagnosis of recurrent/refractory disease, with the latter defined as suboptimal response to initial chemotherapy and/or persistent disease after therapy completion; if the parent spoke English or Spanish and was aged >18 years; and if the child was aged <18 years. Parents were approached consecutively in person for an informed consent discussion and given a study letter. To acknowledge parents’ time and effort, a $50 gift card was offered at the time of survey completion. Medical records dating from the time of diagnosis and consenting parents’ contact information were forwarded to the lead site, Boston Children’s Hospital/Dana-Farber Cancer Institute via the secure Research Electronic Data Capture (REDCap) database.

TABLE 1.

Parent, Child, and Care Characteristics (N = 95)

Parent Characteristics	Percentagea
Sex
Female	81
Male	19
Education
>High school	82
≤High school	18
Race/ethnicity
White	58
Black	8
Hispanic	23
Other	11
Site
Boston Children’s Hospital/Dana-Farber Cancer Institute	22
Children’s Hospital Los Angeles	18
Children’s Hospital of Philadelphia	20
Columbia University Irving Medical Center	3
Cook Children’s Medical Center	7
Seattle Children’s Research Institute	6
St. Jude Children’s Research Hospital	6
Lucile Packard Children’s Hospital at Stanford	9
University of Chicago	7
Care at study site
Primary oncology care	64
Second opinion	36
Child Characteristics
Sex
Female	40
Male	60
Median age at parent enrollment (IQR), y	6 (4–8)
Care and Treatment Characteristics
Neuroblastoma status
Recurrent	76
Refractory to initial therapy	23
Median mo between diagnosis and first recurrence (IQR)	13 (5–24)
Median mo between first recurrence and survey (IQR)	11 (5–25)
Median no. of prior chemotherapeutic regimens (IQR)	3 (2–5)
Median no. of prior clinical trials (IQR)	1 (1–2)
Documented treatment options	80
Experimental therapy on a phase 1/2 clinical trial	58
Established high-intensity regimen	41
Established low-intensity regimen	26
No cancer-directed therapy	12
Treatment decision
Experimental therapy on a phase 1/2 clinical trial	43
Established high-intensity regimen	27
Established low-intensity regimen	26
No cancer-directed therapy	5
Documented prognosis discussion since last recurrence	33
Medical care since last recurrence
>5 hospital d/mo	26
>5 outpatient or emergency department visits/mo	8
ICU care	13
Mechanical ventilation	1
Blood pressure support	1
Bacteremia	9
Highly myelosuppressive therapy requiring stem cell infusion	18
Composite outcomes
Any of the above	49
Frequent inpatient or outpatient visitsb	29
Intensive medical carec	19
Care and Treatment Characteristics
Palliative care discussion	7
Hospice use	2
DNR order or home comfort care form	0

Abbreviations: DNR, do not resuscitate; ICU, intensive care unit, IQR, inter-quartile range.

^aShown as the percentage unless otherwise specified.

^bMore than 5 hospital days or >5 outpatient/emergency department visits per 30 days.

^cAny ICU care, mechanical ventilation, blood pressure support, or bacteremia.

Trained research staff at Boston Children’s Hospital/Dana-Farber Cancer Institute administered questionnaires in English or Spanish to consenting parents; staff at one site (Cook Children’s Medical Center) administered the questionnaires themselves. When possible, questionnaires were administered verbally, either by telephone or in person; paper or electronic versions were offered as an alternative. Longitudinal survey data also were collected but were not presented herein.

Questionnaire content was developed based on our theoretical model using closed-ended items with Likert-type responses from previously developed and validated surveys, with minor modifications when needed, and with review performed by bereaved parents of children with neuroblastoma. Questionnaires were translated into Spanish, back-translated into English, and reviewed for accuracy by 2 bilingual interviewers.

Medical records were reviewed by trained study team members using a detailed algorithm,^15,16 with a secondary reviewer (J.W.M.) reviewing select data elements such as prior treatment regimens.

Parents’ understanding of prognosis

Parents’ understanding of prognosis was assessed by asking, “How likely do you think it is that your child will be cured of neuroblastoma?”^9,17,18 Table 2 shows the response categories.

TABLE 2.

Parent Understanding of Prognosis and Goals of Care

	Percentage
Parent-Reported Likelihood of Cure (N = 91)
Extremely likely (>90% chance of cure)	24
Very likely (75%−90% chance of cure)	13
Moderately likely (50%−74% chance of cure)	19
Somewhat likely (25%−49% chance of cure)	18
Unlikely (10%−24% chance of cure)	5
Very unlikely (<10% chance of cure)	15
No chance of cure	5
Parent’s Single Most Important Goal of Care (N = 94)
Cure	72
Longer life for the child	10
Quality of life	18

Goals of care

Parents were asked, “Parents have many goals of care for their children with neuroblastoma. What are your goals for your child’s care right now?” Parents could choose all applicable goals, with responses of “to cure my child’s cancer,” “to help my child live as long as possible,” and “to relieve pain and discomfort, and improve quality of life, as much as possible.” Parents were then asked, “If you had to choose one, what would you say was your most important goal of your child’s care right now? Many of these goals may be important to you, but please choose the one goal that you feel is most important to you right now,” with identical responses.^3,18

Parents also were asked to respond to hypothetical scenarios^19–21 incorporating the concept of the “good parent” role as a compass for decision making.^22,23 The survey explained, “Sometimes when parents make decisions for their children, they think about what a good parent would do in their situation. But being a good parent can mean different things to different people.” The first scenario asked, “I would like for you to imagine what a good parent might do in another situation. Imagine that doctors have told this parent that her child has no realistic chance for cure, no matter what treatment the child is given. The child is feeling well, able to play and take part in normal activities. What do you think a good parent might do in this situation?” Response options matched those provided for parents’ own goals (focus on cure, longer life, and quality of life). In the second scenario, parents were told, “this time, the child is not feeling well–she has some discomfort, sleeps much of the time, and is not eating or drinking very much,” with the same response options.

Experiences with prior care

Parents were asked whether their child experienced pain, dyspnea, fatigue, poor appetite, nausea and/or vomiting, constipation, or diarrhea within the past 3 months,³ and whether the child suffered from the symptom “a great deal,” “a lot,” “somewhat,” “a little,” or “not at all.”³ Treatment-related benefits and suffering were identified by asking parents to what extent they believed the most recent cancer treatment resulted in benefit (or suffering) (“a great deal,” “a lot,” “somewhat,” “a little,” or “not at all”).¹⁹ Medical records identified markers of higher intensity medical care, including ICU care, mechanical ventilation, blood pressure support, bacteremia, or highly myelosuppressive therapy requiring stem cell infusion, and the number of prior chemotherapeutic regimens and clinical trials.

Care decisions

Care decisions were established using medical records. Two study investigators (J.W.M. and S.S.) classified treatment regimens as clinical trials; established high-intensity regimens, with significant expected toxicity; established low-intensity regimens, with limited expected toxicity; and no cancer-directed therapy and/or supportive care only. Medical records also identified hospice use and do-not-resuscitate orders.

Child quality of life

Child quality of life was assessed using the Pediatric Quality of Life Inventory (PedsQL)^24–26 cancer summary score.

Parental regret

Parental regret concerning the most recent treatment decision was assessed using the Decision Regret Scale^27–29 using a prespecified cutoff value.³⁰ Parents responded to 5 statements: 1) “I have made the right decisions”; 2) “I regret the choices that were made”; 3) “I would make the same choices if I had to do it all over again”; 4) “the choices caused harm”; and 5) “the decisions were wise.” Response options were “strongly agree,” “agree,” “neither agree nor disagree,” “disagree,” or “strongly disagree.”

Parent factors

Parent factors included self-reported race/ethnicity, sex, age, educational level, and marital status. Medical records identified any documented discussions regarding prognosis and palliative care.

The institutional review boards of the participating sites approved the current study.

Statistical Methods

Chi-square tests evaluated differences in proportions or distributions of categorical responses between groups. Agreement between paired items was evaluated using the kappa statistic, and marginal homogeneity was examined using McNemar-Bowker tests for symmetry.³¹ Factors associated with quality of life (PedsQL score >the median) and regret were evaluated using bivariate logistic regression. The small sample size precluded multivariable modeling.

RESULTS

Of the 152 parents approached, 117 (77%) agreed to participate (27 parents declined, 7 neither consented nor declined, and 1 child died after the parent was approached but before a decision regarding enrollment was made). A total of 95 parents (81% of consenting parents, 63% of those approached) completed questionnaires; 8 children died before completion of the parental survey, 2 parents withdrew consent, and 12 parents never completed the survey. Participants did not differ significantly from non-participants with regard to race/ethnicity (P = .55) or their child’s sex (P = .59).

The majority of study participants were mothers (Table 1). Approximately 58% were white, 8% were black, 23% were Hispanic, and 11% were of other race/ethnicity. The median child age was 6 years. The majority of children had recurrent neuroblastoma (76%); 23% had refractory disease. Children had received a median of 3 prior chemotherapeutic regimens (interquartile range, 2–5 regimens) and were enrolled on 1 clinical trial (interquartile range, 1–2 clinical trials).

Treatment options and care received

At the time of the most recent disease recurrence, approximately 80% of records included documentation of a discussion regarding treatment options. Documented options included enrollment on a phase 1/2 clinical trial (documented for 58% of patients), treatment with an established high-intensity regimen (documented for 41% of patients), treatment with an established low-intensity regimen (documented for 26% of patients), and no cancer-directed therapy (documented for 12% of patients). Overall, approximately 43% of parents enrolled their child on a clinical trial after the most recent disease recurrence; 27% chose a high-intensity, nonresearch regimen; 26% chose a low-intensity, nonresearch regimen; and 5% chose no cancer-directed therapy. Discussions regarding prognosis (33%) and palliative care (7%) were documented for a minority of children. Since the time of the most recent disease recurrence, nearly one-fifth of children (19%) received high-intensity medical care, including ICU care (13%), treatment of bacteremia (9%), or stem cell infusion (18%). Approximately 2% of children were enrolled in hospice care; none of the children had do-not-resuscitate orders.

Parent expectations for cure

The majority of parents markedly overestimated their child’s chances of cure (Table 2), with 24% of parents reporting that their child had a >90% likelihood of cure. Approximately 13% of parents reported chances of cure of 75% to 90%, and 19% reported chances of cure of 50% to 74%. Overall, only 26% of parents recognized the poor prognosis, defined as a chance of cure of <25%. Perceptions of prognosis did not differ significantly between the parents of children with recurrent or refractory disease (P = .42); between those parents of children seen for second opinions versus parents of those receiving ongoing oncology care at study sites (P = .32); or by parent sex (P = .47), educational level (P = .77), time from diagnosis (P = .77), study site (P = .91), or method of survey completion (P = .10).

Prior experiences with care

Greater than two-thirds of parents (69%) attributed a great deal or a lot of benefit to the last treatment before study enrollment, and approximately 36% attributed a great deal or a lot of suffering to the last treatment. Approximately 14% of parents reported a great deal or a lot of suffering without a great deal or a lot of benefit. Reports of benefit (P = .64) and suffering (P = .23) did not differ by regimen type. Parents reported high symptom burden, including pain (76%), poor appetite (73%), nausea and/or vomiting (72%), fatigue (67%), diarrhea (63%), constipation (38%), and shortness of breath (17%). Approximately 31% of parents reported that their child experienced a great deal or a lot of suffering from pain within the past 3 months, and 59% reported their child experiencing a great deal or a lot of suffering from any symptom.

Goals of care

When asked to identify all current goals of care, parents placed value on multiple goals, with 92% holding goals of cure, 79% holding goals of a longer life, and 64% holding goals of quality of life. When asked to choose a single most important goal, 72% chose cure, 10% chose longer life, and 18% chose quality of life (Table 2). Care goals were found to be associated with prognostic awareness (P = .002) (Fig. 2A), but not suffering from symptoms (P = .69) (Fig. 2B).

Figure 2.

(A) Relationship between parent understanding of prognosis and goals of care among 90 parents with complete responses to both items. The P value was calculated using the chi-square test. (B) Relationship between parent report of suffering from symptoms and goals of care among 90 parents with complete responses to both items. The P value was calculated using the chi-square test. (C) Parents’ goals of care for their own child and goals of care they believe a “good parent” would have for a hypothetical child among 90 parents with complete responses to all 3 items.

* P values are calculated using tests of symmetry in comparison with parent's primary goal of treatment for his or her own child.

When asked to consider the goal of a hypothetical “good” parent for a child who had no realistic chance of cure but was feeling well, approximately 48% of parents said a good parent should prioritize quality of life (P < .001 compared with goals for their own child) (Fig. 2C). For a hypothetical child with no realistic chance of cure who was not feeling well, approximately 82% of parents said a good parent should prioritize quality of life (P < .001 compared with goals for their own child) (Fig. 2C) (see Supporting Table 1 for complete data).

Factors associated with child quality of life and parental regret

The child’s quality of life (PedsQL score >the median) was found to be inversely associated with parent reports of suffering with limited benefit from the last treatment (odds ratio [OR], 0.11; 95% CI, 0.01–0.87), suffering from pain (OR, 0.11; 95% CI, 0.03–0.36), and suffering from any symptom (OR, 0.11; 95% CI, 0.04–0.29) (Table 3). However, quality of life was not found to be significantly higher when parents reported that quality of life was the primary goal of care.

TABLE 3.

Marginal Association Between Factors and Care Outcomes

	Care Outcome
	QOL (PedsQL Score >Median)		Parental Regret
Factor	Percentage With High QOL	OR (95% CI)a	Percentage With Regret	OR (95% CI)a
Overall	50		41
Parent characteristics
Male sex	53	1.19 (0.41–3.43)	50	1.09 (0.36–3.32)
Education ≤high school	54	1.20 (0.37–3.89)	56	1.81 (0.63–5.27)
Race/ethnicity
White	44	Reference	45	Reference
Black	50	1.26 (0.28–5.60)	13	0.17 (0.02–1.50)
Hispanic	67	2.52 (0.87–7.27)	33	0.60 (0.21–1.74)
Other	44	1.01 (0.24–4.19)	56	1.51 (0.36–6.26)
Understanding of prognosis
Recognition of poor prognosis	43	0.66 (0.25–1.72)	46	1.41 (0.55–3.64)
Care experiences
>Median mo between diagnosis and survey	52	1.73 (0.52–5.76)	40	0.77 (0.24–2.51)
>Median prior treatment regimens	52	1.20 (0.52–2.74)	50	2.06 (0.88–4.83)
High medical intensity of care	48	0.84 (0.37–1.92)	55	3.14 (1.31–7.51)
Last treatment regimen caused suffering with limited benefit	10	0.11 (0.01–0.87)	73	4.78 (1.16–19.72)
A great deal/a lot of suffering from pain	16	0.11 (0.03–0.36)	52	1.92 (0.77–4.78)
A great deal/a lot of suffering from any symptom	28	0.11 (0.04–0.29)	51	2.91 (1.18–7.16)
Goals of care
QOL goal	35	0.46 (0.15–1.38)	24	0.39 (0.12–1.33)
Care decisions
Decision for low-intensity or no cancer-directed therapy	60	1.61 (0.62–4.19)	44	1.35 (0.52–3.50)
Care outcomes
PedsQL score >median	NA	NA	31	0.52 (0.22–1.24)

Abbreviations: NA, not applicable; OR, odds ratio; PedsQL, Pediatric Quality of Life Inventory; QOL, quality of life.

^aBold type indicates P < .05.

Approximately 41% of parents expressed regret concerning the most recent treatment decision. Parents were more likely to experience regret if the child had received higher intensity care (OR, 3.14; 95% CI, 1.31–7.51), experienced suffering with limited benefit from the last treatment (OR, 4.78; 95% CI, 1.16–19.72), or experienced suffering from symptoms (OR, 2.91; 95% CI, 1.18–7.16).

DISCUSSION

Herein, we sought to understand drivers of decision making among parents of children with poor prognoses and a range of therapeutic options. In this study of parents confronting a childhood condition for which cure rarely is achieved, just greater than 25% of parents recognized that their child was unlikely to be cured. Cure was the most commonly endorsed primary goal of therapy, with fewer than 20% of parents found to be focused primarily on quality of life. Consistent with our hypotheses, prior retrospective data,¹⁰ and literature regarding adults,¹⁸ parents who recognized a poor prognosis were much more likely to prioritize quality of life over other goals such as cure.

Previous work in pediatric oncology has shown that bereaved parents who recalled recognizing their child’s poor prognosis earlier had earlier hospice enrollment and earlier institution of do-not-resuscitate orders.¹⁰ However, this study was limited by retrospective parental recall of the timing of this realization, with the potential for recall bias.¹⁰ Because so many pediatric cancers have good prognoses, work regarding prognosis communication in children with advanced pediatric cancer has been limited by the challenges of identifying sufficient numbers of children with uniformly poor prognoses whose parents could be studied prospectively. We addressed these challenges by studying children with a single disease, recurrent or refractory neuroblastoma, as a model for all poor- prognosis pediatric cancers. Doing so required a collaboration between 9 pediatric cancer centers to recruit a sufficient number of parents of children with this rare condition, but this approach also allowed us to identify significant gaps in prognostic understanding.

Prioritizing quality of life over cure for one’s child, or even expressing true prognostic awareness, might create complex conflicts for parents. Therefore, vignettes also were used to explore the ability of parents to prioritize quality of life for children with a poor prognosis. Prior literature has suggested that asking parents about their beliefs regarding the role of a hypothetical “good parent” may be one way to identify underlying values for care,^22,23 and the current study findings support this concept. Although approximately 18% of parents prioritized quality of life for their own children, nearly one-half reported they would do so for a hypothetical child with no realistic chance of cure who was feeling well, and greater than 80% said they would do so if the child was not feeling well.

It is interesting to note that we considered parent reports of a <25% chance of cure to indicate prognostic awareness; although the majority of children in the current study had incurable cancer, children with refractory disease have some small prospect of cure. This differs from language in vignettes and may underlie differences in vignette-based priorities compared with goals for their own children. However, the shift in priorities as evidenced by vignettes also raises concern that parental decision making may not reflect parents’ values; instead, in many cases, priorities expressed for their own children may reflect misperceptions of the child’s ultimate prognosis.

Contrary to our hypothesis, although parents’ responses were influenced by treatment burden and suffering in vignettes, their own children’s prior experiences with suffering did not appear to influence actual care goals. Parents may have been willing to accept significant suffering for their child in the hope of cure. This perspective may be understandable given that many parents believed cure was possible. However, children’s experiences with symptoms, suffering, and medically intensive care are associated with parental regret, raising concern about parental decision making as parents reflect on care.

The findings of the current study underscore the central importance of prognostic communication. Parents are not always able to absorb information regarding prognosis, even when stated clearly, and the uncertain trajectory of illness for any one child adds complexity to prognostication. However, although clinician communication is likely not the only factor at work in this case, it is one over which clinicians have agency. The current study findings have argued strongly for clear communication regarding a child’s poor prognosis so that parents can formulate goals of care that match their personal values.

Because the current study examined patients with a rare condition, the sample size was small, thereby limiting our ability to detect small effects. In addition, similar to other studies of parental decision making, the majority of participants were mothers,^9,30,32 and while nearly 25% of parents were Hispanic, black parents were underrepresented. Study sites predominantly were larger pediatric cancer centers, and parents seeking clinical trials may have been overrepresented. Although the number of documented discussions regarding prognosis and palliative care was strikingly low, some conversations may not have been documented. Parents were recruited at any time during the disease trajectory after the occurrence of recurrent/refractory disease, thereby limiting our ability to compare quality of life and symptoms across the sample. Finally, the current study did not assess clinician expectations of prognosis, and we do not know why parents reported prognosis as they did. We hypothesized that communication regarding prognosis was inadequate or ineffective, but it is possible that some parents recognized a poor prognosis while believing their child would be exceptional, or reported hopes rather than expectations. However, the findings of the current study point to understanding of prognosis as a central underlying driver of parental decision making. We believe that, despite its limitations, the current prospective study in patients with a uniformly poor-prognosis condition has potential implications for all pediatric life-threatening illnesses.

Children with advanced cancer experience significant symptoms and suffering. In the current study cohort of children with poor prognoses and a wide range of care options, few parents were found to recognize poor prognosis, and few identified quality of life as the dominant priority over other goals. Although we had hypothesized that parents learn from experience and shift treatment goals toward palliation in response to prior burdensome experiences, the findings herein have suggested that parents will choose intensive therapies unless they understand their child’s prognosis. Efforts to develop effective prognosis communication methods must be a priority for the field.