Wolfe J, Klar N, Grier HE, Duncan J, Salem-Schatz S, Emanuel EJ, Weeks JC. Understanding of Prognosis Among Parents of Children Who Died of CancerImpact on Treatment Goals and Integration of Palliative Care. JAMA. 2000;284(19):2469–2475. doi:10.1001/jama.284.19.2469
Author Affiliations: Departments of Pediatric Oncology (Drs Wolfe and Grier), Biostatistical Science (Dr Klar), and Adult Oncology (Dr Weeks), and the Center for Outcomes and Policy Research (Drs Wolfe and Weeks), Dana-Farber Cancer Institute, Boston, Mass; the Department of Medicine (Drs Wolfe, Grier, and Salem-Schatz, and Ms Duncan), Children's Hospital, Boston, Mass; Department of Medicine, Brigham and Women's Hospital, Boston, Mass (Dr Weeks); Division of Preventive Oncology Cancer Care Ontario, Toronto (Dr Klar); and Department of Clinical Bioethics, National Institutes of Health, Bethesda, Md (Dr Emanuel).
Context Parents' understanding of prognosis or decision making about palliative
care for children who die of cancer is largely unknown. However, a more accurate
understanding of prognosis could alter treatment goals and expectations and
lead to more effective care.
Objectives To evaluate parental understanding of prognosis in children who die
of cancer and to assess the association of this factor with treatment goals
and the palliative care received by children.
Design, Setting, and Participants Survey, conducted between September 1997 and August 1998, of 103 parents
of children who received treatment at the Dana-Farber Cancer Institute and
Children's Hospital, Boston, Mass, and who died of cancer between 1990 and
1997 (72% of those eligible and those located) and 42 pediatric oncologists.
Main Outcome Measure Timing of parental understanding that the child had no realistic chance
for cure compared with the timing of physician understanding of this prognosis,
as documented in the medical record.
Results Parents first recognized that the child had no realistic chance for
cure a mean (SD) of 106 (150) days before the child's death, while physician
recognition occurred earlier at 206 (330) days before death. Among children
who died of progressive disease, the group characterized by earlier recognition
of this prognosis by both parents and physicians had earlier discussions of
hospice care (odds ratio [OR], 1.03; 95% confidence interval [CI], 1.01-1.06; P = .01), better parental ratings of the quality of home
care (OR, 3.31; 95% CI, 1.15-9.54; P = .03), earlier
institution of a do-not-resuscitate order (OR, 1.03; 95% CI, 1.00-1.06; P = .02), less use of cancer-directed therapy during the
last month of life (OR, 2.80; 95% CI, 1.05-7.50; P
= .04), and higher likelihood that the goal of cancer-directed therapy identified
by both physician and parent was to lessen suffering (OR, 5.17; 95% CI, 1.86-14.4; P = .002 for physician and OR, 6.56; 95% CI, 1.54-27.86; P = .01 for parents).
Conclusion Considerable delay exists in parental recognition that children have
no realistic chance for cure, but earlier recognition of this prognosis by
both physicians and parents is associated with a stronger emphasis on treatment
directed at lessening suffering and greater integration of palliative care.
Cancer remains the leading cause of nonaccidental death in childhood.1 We previously reported that, according to their parents,
children who die of cancer experience significant suffering and inadequate
palliation in the last month of life.2 One
potential obstacle to more effective integration of palliative care into the
treatment of children with advanced cancer is that unrealistic physician and/or
parental prognostic expectations may be leading to inappropriate treatment
goals. Yet little is known about the understanding of prognosis and treatment
preferences of parents of children who die of cancer. Studies in adults with
terminal cancer have demonstrated that communication between physicians and
patients about treatment preferences is suboptimal.3- 9
A number of studies have shown that adults with cancer tend to overestimate
their survival probabilities, and these unrealistic expectations are, in turn,
related to preferences for more aggressive therapies.3,7,9
It is not known if the experiences among dying children are similar.
In children, communication about end-of-life issues may be more challenging
because the parent is almost always the surrogate decision maker.10- 12 Saunders13 has validated the clinical impression that parental
grief is more severe and longer lasting than other types of grief. The anticipation
of losing a child may impede discussions about the terminal prognosis and,
in turn, may impact decision making and integration of palliative care.14 It also is possible that physicians of children with
advanced cancer have difficulty confronting the loss of a young patient since
the intent in pediatric cancer care is generally curative.15
As such, greater barriers may exist to optimal communication about end-of-life
issues in children with advanced cancer and may result in poorer quality-of-life
outcomes for the 25% of children who will eventually die of their disease.1
We interviewed parents, surveyed the primary oncologist, and reviewed
the medical records of children who died of cancer to answer the following
questions: When do parents and physicians become aware that the child has
no realistic chance for cure? What factors are associated with a greater concordance
in timing of parent and physician awareness that the child has no realistic
chance for cure? Does understanding of the child's prognosis alter parents'
treatment goals? What outcomes are associated with earlier parent-physician
recognition that the child has no realistic chance for cure?
As previously described,2 the study was
conducted at Children's Hospital and the Dana-Farber Cancer Institute, Boston,
Subjects included parents of children who died of cancer between 1990
and 1997. Parents were eligible if they were (1) English-speaking, (2) residents
of North America, (3) their child had died of cancer more than 1 year before
enrollment, and (4) the child's former physician permitted contact with the
family (permission was denied for 15 families). Eligible parents were sent
a letter containing a postage-paid "opt-out" postcard. Of 165 eligible parents,
143 were located, and 107 agreed to participate. We conducted a single interview
of 1 parent per family; which parent participated was left to the family's
discretion. One hundred three parents were interviewed between September 1997
and August 1998 for a response rate of 72% (103/143). The mean duration of
the interview was 113 minutes. The parent interview was conducted a mean (SD)
of 3.1 (1.6) (range, 1.1-8.0) years following the death of the child. We also
surveyed the child's primary oncologist and reviewed medical charts.
Relevant domains and draft items were developed from literature review
and focus groups of parents and medical caregivers.2
Whenever possible, items were taken from previously validated surveys.16 However, the majority of items were newly created
following guidelines suggested by Streiner and Norman.17
All items were closed-ended with categorical responses or Likert scales. A
pretest of the instrument assessed content, wording, response burden, cognitive
validity, and parental willingness to participate.
A trained interviewer and 1 of the investigators (J.W.) conducted the
parent interviews. The majority of interviews were administered by telephone;
5 were conducted in person at the request of the parent. Parents were asked
to estimate the timing of the onset of their understanding that the child
had no realistic chance for cure, also defined as the end-of-life care period.
They were asked if they had ever discussed this prognosis with medical caregivers,
and if so, who participated in this discussion, whether the topic was appropriate
to discuss, and how comfortable they were with the manner in which this was
discussed. Parents also reported their understanding of the primary goal of
cancer-directed therapy at the time of diagnosis, during the period when they
understood that the child had no realistic chance for cure, and what they
currently believed the primary goal of cancer-directed therapy should be during
the end-of-life care period for children with cancer. Response categories
included "to cure your child's cancer," "for you and/or your child to be able
to keep hoping," "to make sure you have done everything," "to extend your
child's life as long as possible but without hope of cure," "to lessen your
child's suffering as much as possible," "to help cancer research," and "other."
Parents were asked about their primary goal of treatment of symptoms during
the end-of-life care period ("to extend life as long as possible," "to lessen
suffering as much as possible," or "other"). Additional data collected in
the parent interview included understanding of the child's likelihood of cure
at the time of diagnosis, level of preparedness for symptoms experienced during
the end-of-life care period, quality of care provided by the primary team,
quality of communication with the primary team, involvement of a home care
team (hospice or visiting nurse association) and/or a psychosocial clinician
(psychologist or social worker) during the end-of-life care period, peacefulness
of the child's death, and parent demographics (age, sex, race, education,
income, and religion). Parents were informed in the introductory letter and
interview that psychosocial support was available to them if desired.
Physicians were asked what they believed to be the child's likelihood
of cure at diagnosis, the primary goal of treatment at 1 month prior to death,
and perceived peacefulness of the death. We determined the number of years
of experience by subtracting the physician's date of beginning internship
from the patient's date of death. The physician's sex also was recorded.
Charts of the children of all eligible parents were abstracted by trained
research assistants. Every tenth chart also was abstracted by one of the study
investigators. When discrepancies were found, the chart was rereviewed by
both abstractors, and the appropriate response was determined by consensus.
The timing of the onset of the end-of-life care period by physician determination
was defined as the first date when chart documentation included a statement
consistent with there being no realistic chance for cure, any reference to
do-not-resuscitate status, or any reference to hospice or home care to provide
terminal care. Additional data collected by chart review included sex; diagnosis;
dates of birth, diagnosis, and death; number, types, and timing of cancer-directed
regimens and enrollment in clinical studies; cause of death classified as
progressive disease with or without a treatment-related complication or a
treatment-related complication alone; intubation in the last 24 hours of life;
use of cardiopulmonary resuscitation; and location of death.
The institutional review boards of the 2 participating institutions
approved the study. All parents gave verbal informed consent.
Analyses were conducted using the SAS statistical package (SAS Institute
Inc, Cary, NC). Respondents who did not answer a specific question or for
whom data were not available from chart review were excluded from the analyses
of that data element. Variables using Likert scales were dichotomized as specified
in text and tables. The κ statistic18
was used to look for agreement between physician and parental views regarding
elements of end-of-life care. The McNemar test18
was used to compare the direction of discordance, when present.
We determined the mean difference between physician and parent estimates
of the onset of the end-of-life care period and looked for factors associated
with this discrepancy using linear regression analyses. We adjusted for physician
clustering to account for the possibility that children might have more similar
data by virtue of sharing the same physician.19
Associations among clinical, sociodemographic, and attitudinal variables were
examined in bivariate analyses. Factors found to be associated with P<.05 were entered into multivariate models.
We also examined factors associated with the parent-physician group
in which there was recognition that the child had no realistic chance for
cure at 50 days prior to death among children who died of progressive disease,
using logistic regression analyses. We selected 50 days because at that point
the physician and parent for half the cohort (51%) recognized that the child
had no realistic chance for cure, and for the other half, the physician, parent,
or both did not believe that the child had entered the end-of-life care period.
Again, we adjusted for physician clustering.19
All models were fit using the SAS Genmod procedure.
The mean (SD) age of parents at the time of the interview was 43 (7.7)
years. Most were white (91%) and women (86%). The range of reported median
household income was $25,000 to $49,999 per year. Seventy-six percent of parents
had some postsecondary education and 50% were Catholic.
The clinical characteristics of the children are summarized in Table 1. More than one third of the cohort
underwent bone marrow transplantation; 5% underwent 2 bone marrow transplantations.
Twenty percent of children died as a result of a treatment-related complication.
When compared with the children of parents who were interviewed, children
of nonrespondents did not differ significantly with regard to sex, diagnosis,
age at death, duration of disease, enrollment in a clinical study, having
undergone bone marrow transplantation, or place of death. Children of parents
who did not participate were more likely to have undergone cardiopulmonary
resuscitation (26% vs 6%, P = .03).
Of the 42 pediatric oncologists who cared for the children, 13 (31%)
were women. They had a mean (SD) of 7.1 (4.7) years of experience at the time
of death of the child.
Fifty-six percent of parents and 43% of physicians reported that they
believed cure was likely at the time of diagnosis (κ = 0.492; 95% confidence
interval [CI], 0.32-0.66). Among physicians and parents whose recalled prognostic
estimates differed, parents were significantly more likely to believe that
the child would be cured (odds ratio, 3.0; 95% CI, 1.2-7.6; P = .02).
For parents, the first recognition that the child had no realistic chance
for cure occurred a mean (SD) of 106 (150) days prior to death. There was
no significant association between estimates of the duration of this period
and the length of time elapsed between the child's death and the parent's
interview. In contrast, the mean (SD) length of time between the first physician
documentation that the child had entered the end-of-life period and the child's
death was 206 (330) days. The mean (SD) of the difference between physician
and parent estimates of the onset of the end-of-life care period for each
child was 101 (314) days. Factors associated with smaller differences between
parent and physician estimates of the onset of the end-of-life care period
are summarized in Table 2. The
differences were smaller if the child had a hematological malignancy (P = .04), if the parent had no more than a high school
education (P = .02), and if a psychosocial clinician
was involved in the end-of-life care (P = .02). These
factors remained significant in multivariate analyses. The difference between
physician and parent estimates of the onset of the end-of-life period was
not statistically significantly associated with any of the end-of-life care
Ninety-five percent of parents reported having had a discussion at some
point about their child having no realistic chance for cure. Seventy-four
percent reported that the primary oncologist participated in this discussion,
27% that a primary nurse was involved, and 30% that a psychosocial clinician
was involved. Only 16% of children participated in this discussion. Ninety
percent of parents felt that it was appropriate to discuss this topic but
only 38% were very comfortable with the manner in which it was discussed.
When asked how they first came to understand that their child had no
realistic chance for cure, only 49% of parents reported that this resulted
from a discussion with the medical team; 30% reported that this understanding
came from a perceived change in the way their child looked or acted; and 9%
reported that this understanding came from a feeling or dream.
Eighty percent of parents reported that at diagnosis, they believed
that the primary goal of cancer-directed therapy was to cure their child (Table 3). During the parent-defined end-of-life
care period, the majority of parents continued to report that the primary
goal of cancer-directed therapy was to extend life, specifically to cure (28%),
to keep hoping (13%), to have done everything (10%), or to extend life without
hope of cure (15%). Only 13% of parents reported that the primary goal of
cancer-directed therapy during this period was to lessen suffering. When asked
what they now think should be the primary goal of cancer-directed therapy
during the end-of-life care of children with cancer, the proportion of parents
who responded to lessen suffering was higher at 34% (odds ratio, 6.5; 95%
CI, 2.3-18.6; P<.001).
In the last month of life, among the 41 children who continued to receive
cancer-directed therapy, 42% of physicians but only 19% of parents reported
that the primary goal of this treatment was to lessen suffering (κ =
0.16; 95% CI, − 0.11-0.42).
Seventy percent of parents reported that they believed that the primary
goal in treating symptoms during the end-of-life care period was to lessen
suffering. Overall, 59% of parents reported that the primary goal of treating
symptoms during this period was to lessen suffering while also reporting that
the primary goal of cancer-directed therapy was to extend life. Furthermore,
when asked what they currently believe should be the primary goal of treatment
of symptoms during the end-of-life care period for children with cancer, the
proportion of parents who responded to lessen suffering was higher at 79%
(odds ratio, 2.8; 95% CI, 1.1-7.2; P = .04).
To assess the relationship between the timing of recognition that the
child had no realistic chance for cure and the outcomes of end-of-life care,
children were classified into groups based on whether parent and physician
recognition occurred more or less than 50 days prior to death. The outcomes
for the group in which there was recognition by both parent and physician
50 days or more prior to death were compared with outcomes for all other children,
adjusted for physician clustering (Table
4). In this group, there was earlier documentation of a discussion
of hospice (P = .01), better parental ratings of
the quality of care delivered by the home care team (P
= .03), earlier institution of a do-not-resuscitate order (P = .02), less use of cancer-directed treatment in the last month of
life (P = .04), and higher likelihood that both physician
and parent identified the primary goal of cancer-directed therapy to be to
lessen suffering (P = .002 for physician and P = .01 for parents).
We sought to evaluate the understanding of prognosis among parents of
children who died of cancer and to assess the relationships between this understanding,
their perceptions of the goals of treatment, and adoption of palliative care
strategies. We found that at diagnosis, when both physicians and parents were
relatively optimistic about the child's prognosis, there was fairly good concordance
between parents' and physicians' beliefs about the likelihood of cure (κ
= 0.492). However, as the children's cancer advanced, parents' understanding
that the child no longer had a realistic chance for cure was delayed, lagging
behind the explicit documentation of this fact by the primary oncologist by
more than 3 months. One explanation for these findings is that communication
about prognosis between physicians and parents at the time of diagnosis may
be clearer than when a child's cancer is more advanced.20
Since 75% of children with cancer will be cured of their disease, pediatric
oncologists have comparatively less experience communicating about terminal
prognoses and may find it difficult to do so. This hypothesis is supported
by the finding that although nearly all parents reported having a discussion
at some point with a medical caregiver about their child having no realistic
chance for cure, only 49% of parents reported that they came to understand
that their child was terminally ill through this discussion.
Greater concordance between the timing of physician and parental understanding
that the child had no realistic chance for cure was not associated with end-of-life
care outcomes. However, when both the physician and parent recognized earlier
that the child had no realistic chance for cure, elements of palliative care
were more likely to be integrated into the child's care. Specifically, hospice
was introduced earlier, and parents were more satisfied with the quality of
home care during the end-of-life care period. Furthermore, the child was less
likely to continue to receive cancer-directed therapy in the last month of
life, and importantly, both the physician and parent were more likely to have
identified the primary goal of treatment to be to lessen suffering. These
outcomes are congruent with recommendations for optimizing care at the end
of life.21- 25
In a cross-sectional retrospective study design such as this, the direction
of the causal relationship between these factors cannot be established with
certainty. For example, discussions of hospice or resuscitation status may
have precipitated, rather than resulted, from parental understanding of the
child's terminal prognosis. Future prospective studies are needed to fully
delineate these relationships. Nonetheless, our findings suggest that interventions
aimed at facilitating earlier recognition by both physicians and parents that
an illness may be fatal could contribute to improving quality of life at the
end of life.
We also found that parents of children with advanced cancer can maintain
dual goals of care concurrently. Specifically, during the parent-defined end-of-life
care period, 59% of parents reported that the primary goal of treating symptoms
was to lessen suffering at the same time that they reported that the primary
goal of cancer-directed therapy was to extend life. These views might be considered
incompatible under the traditional model of palliative care that involves
a definitive transition from curative to supportive care. Our findings indicate
that parents may not have to fully acknowledge their child's poor prognosis
to be willing to emphasize lessening of suffering. There is a growing consensus
that in patients with life-threatening illness, palliative care should be
integrated early and concurrent with treatment of the underlying disease.26 Parents may be comfortable with such an approach,
and our findings suggest a framework in which caregivers could broach the
topic of integrating palliative care earlier while not eroding hope.
One of the more intriguing findings of our study was that when a psychosocial
clinician (psychologist or social worker) was involved in the child's care
there was greater concordance in the timing of physicians' and parents' understanding
that the child had no realistic chance for cure. Further study to determine
whether an interdisciplinary approach to care of children with advanced disease
enhances communication of outcomes may be warranted. However, the failure
of the Study to Understand Prognoses and Preferences for Outcomes and Risks
of Treatments to demonstrate a benefit in end-of-life care outcomes from involvement
with a specially trained nurse is a cautionary lesson.6
While our data suggest that interventions that focus on strengthening interdisciplinary
interactions could be one way to enhance communication around palliative care,
rigorous evaluation is clearly needed.27- 29
Our study has a number of limitations. The parent-defined onset of the
end-of-life care period was determined an average of 3.1 years after the child's
death and may not accurately reflect their perceptions at the time of their
child's actual experience. However, estimates of the duration of the end-of-life
care period did not vary with the amount of time that had elapsed from the
child's death to the parent's interview. Furthermore, every parent was able
to provide information on the timing of the onset of this end-of-life period.
Perhaps coming to understand that one's child has no realistic chance for
cure is such a significant event that it is readily recalled. It is also important
to note that physician estimates of the onset of the end-of-life care period
were taken from the chart documentation at the time of the illness, which
we believed was a more accurate way to estimate when this recognition occurred
than to rely on physician recall. It is not possible to know for certain whether
this accurately reflects the true onset of physician understanding of the
child's prognosis or whether the timing of documentation coincided with the
timing of the actual discussion with parents. At a minimum, however, the timing
of this documentation reflects when effective communication should have taken
place, and it appears that this did not occur, whether due to physician factors,
parental factors, or both. Future prospective studies are needed to more fully
characterize the dynamics of this process.
Selection bias also may have influenced our study findings. Physicians
denied us permission to contact 8% of families, and 22% of parents declined
to participate. The only characteristic that differed between children of
respondents and nonrespondents was the proportion who underwent cardiopulmonary
resuscitation. This is a significant end-of-life event and it is not possible
to know whether findings would have differed in this group. Importantly, we
studied the patterns of care at a single institution. However, the care of
pediatric oncology patients is often regionalized because of the need for
subspecialty trained personnel.30 Thus, the
findings from this 1 large referral center may reflect experiences of most
children with cancer.
Finally, our conclusions about causality among the associations demonstrated
in this cross-sectional study must be made with caution. It is possible that
parents who are more willing to involve a psychosocial clinician in the care
of their child may also be better able to hear and understand what physicians
are saying about the child's prognosis. Furthermore, unidentified parental
factors may enable certain families to both accept prognostic information
and integrate palliation more readily. However, our findings provide preliminary
support for the possibility that the quality of caregiver communication played
an important role in the integration of palliative care.
This study represents one of the first efforts to evaluate the quality
of communication between physicians and parents of terminally ill children
regarding prognosis, treatment goals, and care at the end of life. We have
demonstrated that there may be significant barriers to effective communication
about these emotionally laden issues. Yet, interventions aimed at enhancing
communication around end-of-life care could result in a greater focus on palliation
and less suffering experienced by children with cancer at the end of life.
The potential benefits of such interventions to patients, their families,
and society could be substantial.