- © 2007 by American Society of Clinical Oncology
Hope and Prognostic Disclosure
- Jennifer W. Mack,
- Joanne Wolfe,
- E. Francis Cook,
- Holcombe E. Grier,
- Paul D. Cleary and
- Jane C. Weeks
- From the Departments of Pediatric Oncology and Medical Oncology and the Center for Outcomes and Policy Research, Dana-Farber Cancer Institute; Department of Medicine, Children's Hospital; Department of Epidemiology, Harvard School of Public Health; Department of Health Care Policy, Harvard Medical School; and Department of Medicine, Brigham and Women's Hospital, Boston, MA
- Address reprint requests to Jennifer W. Mack, MD, MPH, Center for Outcomes and Policy Research, Dana-Farber Cancer Institute, 44 Binney St, Boston, MA 02115; e-mail: Jennifer_mack{at}dfci.harvard.edu
Abstract
Purpose Physicians sometimes selectively convey prognostic information to support patients’ hopes. However, the relationship between prognostic disclosure and hope is not known.
Patients and Methods We surveyed 194 parents of children with cancer (overall response rate, 70%) in their first year of treatment at the Dana-Farber Cancer Institute and Children's Hospital (Boston, MA), and we surveyed the children's physicians. We evaluated relationships between parental recall of prognostic disclosure by the physician and possible outcomes, including hope, trust, and emotional distress. Our main outcome was assessed by asking parents how often the way the child's oncologist communicated with them about the children's cancers made them feel hopeful.
Results Nearly half of parents reported that physician communication always made them feel hopeful. Parents who reported receiving a greater number of elements of prognostic disclosure were more likely to report communication-related hope (odds ratio [OR], 1.77 per element of disclosure; P = .001), even when the likelihood of a cure was low (OR, 5.98 per element of disclosure with likelihood of a cure < 25%; P = .03). In a multivariable model, parents were more likely to report that physician communication always made them feel hopeful when they also reported receipt of more elements of prognostic disclosure (OR, 1.60; P = .03) and high-quality communication (OR, 6.58; P < .0001). Communication-related hope was inversely associated with the child's likelihood of cure (OR, 0.65; P = .005).
Conclusion Although physicians sometimes limit prognostic information to preserve hope, we found no evidence that prognostic disclosure makes parents less hopeful. Instead, disclosure of prognosis by the physician can support hope, even when the prognosis is poor.
INTRODUCTION
Physicians and patients alike believe that the best medical communication allows for hope,1-8 no matter how difficult the situation. But this ideal presents challenges for physicians, who may view disclosure of a poor prognosis and provision of hope as conflicting obligations.1,9-17 In service of hope, some physicians therefore choose to limit or carefully tailor the information they give to patients. For example, physicians may give no prognostic information and avoid such a discussion when asked directly,2,18-20 may give overly optimistic information,2,19,20 or may collude in hopes of a cure by focusing on treatment options.9
These choices derive in part from a definition of hope as the belief that a cure is possible. However, the meaning of hope may not be tied to a cure in every situation11,21,22; instead, patients with realistic perceptions of prognosis can transform hopes to reasonable possibilities, such as a meaningful end-of-life period.23-27 Without information that allows for realistic expectations, however, patients may not have the opportunity to formulate such hopes.
A few studies suggest that information about the diagnosis and a willingness on the part of physicians to answer questions can support patients’ senses of hope.5,28-31 The degree to which disclosure of prognosis supports hope is unknown. We sought to evaluate the relationship between hope and prognostic disclosure among parents of children with cancer.
We measured disclosure by asking parents to recall what they had been told about the prognosis by physicians, including whether prognostic information was provided in any form, in quantitative terms, and in written form; whether the physician gave prognostic information before the parent asked; and whether parents wanted additional prognostic information beyond what they had already received. We measured hope by asking parents how often the way the children's oncologists communicated with them about the children's cancers made them feel hopeful. We developed the concept of hope derived from physician communication in response to patients’ assertion in the literature that physician communication should be honest and also should allow for hope. We chose not to tie this concept to a specific medical outcome, such as a cure, because hope for a cure may not be appropriate in every situation. A caring relationship with the physician, however, might in itself be a source of hope. In this way, hope that relates to the communication process might be a goal for every communication interaction. We also evaluated other potential outcomes of prognostic disclosure, including trust and emotional distress, to consider additional possible consequences of disclosure.
PATIENTS AND METHODS
We conducted a cross-sectional, questionnaire-based study of parents and physicians of children with cancer. Some findings have been reported previously.32,33
Study Population
We surveyed parents and physicians of children undergoing treatment for cancer at the Dana-Farber Cancer Institute and Children's Hospital (Boston, MA) between April 2004 and September 2005.32,33 Parents were recruited within 1 year of their child's diagnosis; no time limit was imposed for completion of the questionnaires. Of the 276 eligible parents, 194 (70%) completed the survey. After the parental survey had been completed, the primary oncology attending physician for each patient was administered the physician survey. Ninety-nine percent (193 of 194) of physician surveys were completed, which represented 20 of 21 physicians surveyed.
Data Collection
Procedures of survey development have been previously described.32,33 The primary outcome of this study was communication-related hope, which was assessed by asking parents how often the way the children's oncologists communicated with them about the children's cancers made them feel hopeful. Response options were always, often, sometimes, rarely, or never. Like other questions designed for this survey, this question was developed based on parental interviews and a literature review. Separate questions assessed hopes for medical outcomes; parents were also asked the degree to which hope for a cure and hope for a good quality of life for the child kept them going; response options were a great deal, some, only a little, not at all, or no hope of this. Conceptual, face, and content validities were established through pilot testing that used individualized, cognitive interviews among parents with iterative revision and retesting.
Additional outcomes assessed included parental trust (ie, asking how much parents trust the oncologist's judgment about their children's medical care and using the response options completely, mostly, somewhat, a little, or not at all, derived from the Trust in Physicians Scale34,35); a sense of meaning (derived from the Functional Assessment of Chronic Illness Therapy–Spiritual Well-Being scale, sense of meaning subscale36); and distress related to prognosis communication (ie, asking parents how upsetting is it to know information about your children's prognoses and using the response options extremely, very, somewhat, a little, or not at all).
Prognostic disclosure was the primary independent variable of interest. Survey questions that pertained to parental recall of prognostic disclosure were selected based on conceptual relevance and were combined into a five-item index (Table 1). Validity of a summary score was determined through additional analyses. Individual elements of disclosure had a limited inter-item correlation (r = 0.08 to 0.36), which suggested variability in disclosure practice. Principal components analysis confirmed that selected questions were defined by a rotated-factor matrix that had an eigenvalue greater than 1. We therefore utilized the summary disclosure index (Cronbach's α, .69).
Elements of Prognostic Disclosure and Scoring System
The likelihood of a cure served as an independent variable. Parents and physicians were each asked how likely they now think it is that the child will be cured of cancer, and response options were extremely likely (> 90% chance of a cure), very likely (75% to 90%), moderately likely (50% to 74%), somewhat likely (25% to 49%), unlikely (10% to 24%), or very unlikely (< 10%).37,38
Additional independent variables of interest were the quality of communication (Cronbach's α, .76), the quality of information (Cronbach's α, .87), social support,39 parental sense of faith,36 and identities of those present for the initial conversation about diagnosis and treatment. Questions in the quality of the communication scale addressed physician sensitivity (ie, asking parents how often the children's oncologists conveys information in a sensitive manner), time for questions (ie, asking parents how often the child's oncologist takes enough time to answer questions), clarity (ie, asking parents how often, when asking questions, answers are understandable), and listening (ie, asking parents how often the way the children's oncologists communicated with them about their children's cancers made them feel listened to). Response options were always, sometimes, rarely, or never. The parental survey asked about the parent's sex, relationship to the child, age, education, marital status, race/ethnicity, and religion. The physician survey included questions about the physician experience and the patient's diagnosis and treatment.
The institutional review board of the Dana-Farber Cancer Institute approved this study.
Statistical Methods
We evaluated associations between prognostic disclosure and outcomes of communication, and we evaluated factors associated with communication-related hope. Analyses were conducted using the SAS statistical package (SAS Institute Inc, Cary, NC).
The disclosure index was analyzed as a sum of the number of elements of disclosure, and it ranged from zero to five (Table 1). For communication, social support, and the sense of faith and meaning scales, individual question responses were summed, and the sums were dichotomized at the sample median for analysis. Based on previous literature that suggests that even one negative communication interaction can be troubling to parents,6,40 responses to the question about communication-related hope were dichotomized such that a parental designation that communication always made them feel hopeful was compared with responses of often, sometimes, rarely, and never.
Bivariable and multivariable relationships with communication-related hope were described by using simple logistic regression with generalized estimating equations to account for the clustering of patients within physicians. We used a backward elimination technique to create the multivariable model; the criteria for entry was P = .10 and for retention was P = .05. Variables considered were parent-rated communication quality, information quality, prognostic disclosure, the presence of the child at the initial discussion of diagnosis and treatment, each parent's social support, the sense of faith and meaning, and the physician-rated likelihood of cure. We also tested an interaction between disclosure and the physician-rated likelihood of cure. Models included the physician-rated prognosis; parental sex, race/ethnicity, and education; time since diagnosis, and diagnosis (ie, hematologic malignancy, brain tumor, or other solid tumor), regardless of the significance of their coefficients. We performed sensitivity analyses by using the parent-rated prognosis and by limiting analyses to parents with accurate perceptions of prognosis.
RESULTS
Parent respondents were predominately female, white, married, and well-educated (Table 2).
Parent and Child Characteristics
Nonrespondents did not differ significantly from respondents with respect to the child's sex, age, or diagnosis, although nonrespondents were slightly more likely than respondents to have children with solid tumors (P = .07).
Parents recalled receiving a median of four elements of disclosure (Table 2). Parents most commonly reported discussing the prognosis with the physician (93%), receiving a numeric estimate of prognosis (69%), and receiving information about the prognosis from the physician without needing to ask (67%). Parents were more likely to recall more extensive prognostic disclosure when the child's prognosis was good (P < .0001).
Nearly half (48%) of parents reported that physician communication always made them feel hopeful; 38% reported it as often, 12% reported it as sometimes, 2% reported it as rarely, and 0% reported it as never. The majority of parents stated that hope for a cure (88%) and for a good quality of life (90%) for the child kept them going a great deal. Parental reports of hope derived from physician communication practices were not strongly associated with their reports of hope for specific medical outcomes; responses correlated weakly with parental hopes for a cure (r = 0.13) and for a good quality of life (r = 0.22.)
In bivariable analyses, parents were more likely to report communication-related hope (ie, communication always made them feel hopeful) when they also recalled increased prognostic disclosure (odds ratio [OR], 1.77 per element of disclosure; P = .001; Table 3). Increased recall of prognostic disclosure was associated with communication-related hope even when the child's likelihood of a cure was less than 25% (OR, 5.98; P = .03; Table 4). Although the relationship was not statistically significant for parents of children whose likelihood of a cure was 25% to 74%, the direction of the effect was similar (OR, 1.32; P = .21). Prognostic disclosure and communication-related hope were also associated for parents of children in the best prognostic category (OR, 1.96; P = .0004).
Bivariable Relationships Between Communication Attributes and Parental Report of Hopefulness
Bivariable Relationships Between Prognostic Disclosure and Parental Hopefulness
In a multivariable model, parents who reported a greater degree of prognostic disclosure were more likely to report that physician communication always made them feel hopeful (OR, 1.60 per element of disclosure; P = .03; Table 5). Parents also were more likely to report communication-related hope when they considered communication by the child's oncologist to be of high quality (OR, 6.58; P < .0001). A low physician-reported likelihood of cure was associated with decreased odds of communication-related hope (OR, 0.65 per category of worsening prognosis; P = .005).
Multivariable Model of Communication Attributes Associated With Parental Hopefulness
Multivariable relationships were similar when the parent-rated prognosis was used instead of the physician-rated prognosis (OR of communication-related hope per element of disclosure, 1.43; P = .04). The relationship between hope and recalled disclosure was also similar when we limited our analyses to parents whose perceptions of prognosis agreed with the physicians’ (OR, 2.09; P = .008).
Parents who recalled an increase of prognostic disclosure by the physician were more likely to report that they trusted the oncologist's judgment completely (OR, 1.32 per element of disclosure; P = .005; Table 6). Such parents also tended to have a greater sense of meaning in their experiences of the child's illness (OR, 1.43; P = .02). Parents who reported greater disclosure were less likely to have found prognostic information extremely or very upsetting (OR, 0.68; P < .0001). Relationships were similar when adjusted for the physician-reported likelihood of a cure.
Bivariable Relationships Between Prognostic Disclosure and Outcomes of Communication
DISCUSSION
We should approach a study on hope with some trepidation. The meaning of hope—its origins, its roles—rests with the individual. Yet, physicians use the idea of hope as a reason to engage in patient care in particular ways, at times choosing to limit the provision of information about poor outcomes.2,9,18-20 We evaluated this practice directly by assessing the relationship between recalled prognostic disclosure and communication-related hope among parents of children with cancer. We found that parents tended to derive more hope from physician communication when they also reported that physicians provided detailed prognostic information. This finding counters popular beliefs about the relationship between hope and prognostic information1,9-17; information about prognosis appears to support at least one form of hope—the hope that parents derive from communication with physicians—rather than undermining it.
Because most children in our study had a likelihood of cure that was 75% or greater, we were concerned that the relationship between hope and disclosure was driven by the large proportion of parents with children in the best prognostic categories. Therefore, we looked for such an effect and found that the positive relationship between recalled disclosure and hope was present even for children with a low likelihood of a cure. Although the effect for the intermediate prognostic category was not statistically significant, the direction of the effect was similar, and, most important, no negative effect of disclosure was found.
There are many good reasons to give parents full information about prognosis. Ethical principles establish a fundamental role of information provision in patient autonomy. In addition, knowledge about prognosis affects the decisions patients and families make.21,24,38,41 Perhaps most important, patients and families want prognostic information.33,42,43 A larger study that includes more parents of children in the poorest prognostic categories would be useful to confirm that hope can be enhanced by prognostic information. In light of the many reasons for prognostic disclosure, however, the lack of a negative effect should cause physicians to rethink the practice of nondisclosure.
Others have found that information about a diagnosis can help patients to manage distress,29-31 and our findings suggest that prognostic information could have similar benefits. In addition to greater communication-related hope, parents who recalled more prognostic disclosure reported increased trust in the physician, an enhanced sense of meaning, and decreased distress about prognosis. These relationships were similar after adjustment for prognosis, which suggests that positive outcomes of disclosure are possible even when the prognosis is poor. Other communication attributes could link disclosure with the outcomes we assessed. Even without the indication of causation, however, our findings suggest that physicians who disclose the prognosis in detail can also promote trust and a sense of meaning and can do so without causing undue distress.
The choice to withhold information may stem from a deep sympathy with the patient,10 and the distress physicians and patients experience in such encounters must be acknowledged.16 However, compassion by the physician should not translate to avoidance of important medical information. Communication-related hope was most likely when parents considered communication to be of high quality, a designation that included sensitivity and the feeling that the physician listened to the parent. Of note, even when we controlled for communication quality and recalled disclosure in the multivariable model, parents of children with poor prognoses were less likely to derive hope from physician communication than were parents of children with better prognoses. We may be able to support hope through honest disclosure of information and through our relationships with patients and families, but the deep sadness and sense of loss parents feel in learning of a poor prognosis cannot, perhaps, be fully effaced.
Our measure of hope was created for our questionnaire. Rather than using hope for a cure as our outcome, we evaluated the hope parents derive from conversations with physicians. We chose this measure because we felt it addressed a central issue—can physicians help parents to feel hopeful even while taking away hope for a particular medical outcome? Previous literature has suggested that hope does not always indicate naiveté about outcomes; instead, meaningful experiences and relationships can serve as a foundation for hope.11,21,23,25,26 Our study does not tell us the content of parental hopes; it does tell us, however, that physician communication has the capacity to foster hope, even when the news is bad. If hope is a desired outcome of medical communication, we would argue that it is this kind of hope and not hope based on unrealistic expectations.
We used parental reports of prognosis communication to measure disclosure, but parental recall of discussions about prognosis could be inaccurate. Although we tried to utilize relatively factual reports of what occurred in conversations, confirmation by physician report or by direct observation would be useful. However, the information a parent takes from the encounter with the physician is arguably the most important outcome of communication. In addition, the relationship between hope and disclosure could be explained by the disclosure of overly optimistic prognoses to parents. However, relationships were similar when we evaluated only parents with accurate perceptions of prognosis and when we controlled for parent-rated prognosis. We believe, therefore, that our findings are not simply a reflection of an overly optimistic disclosure.
Additional potential limitations should be considered. This study was cross-sectional, and parents were recruited up to 1 year after the child's diagnosis. Adjustment for the time since diagnosis in our multivariable model did not change our results, though. Because of generally poorer prognoses among children with solid tumors, the lower rate of participation among parents of children with solid tumors raises the possibility that our sample under-represents parents of children with poor prognoses. Future studies should include efforts to recruit parents of such children.
Because we studied parents at a single center that had limited diversity, additional study in a broader population remains important, particularly in consideration of cultural differences in communication preferences.44-48 However, most children with cancer receive care in similar academic centers49; our results, therefore, may apply to most children with cancer. Application to adult patients also is unclear. However, like parents,33 adult cancer patients want prognostic information43 and feel that information provision in general supports hope.5,8,28-31 It seems likely, therefore, that our findings may apply to adults.
We began this discussion with a note of caution: hope is complex, and its meanings are personal. However, certain assumptions have influenced practice—in particular, the physician belief that a discussion of prognosis may disrupt hope. There are many good reasons to be honest with parents about a child's prognosis; honest disclosure supports patient autonomy, the parental desire for information, and the parents’ abilities to make appropriate decisions for their children. Our inability to document harm from the parents’ recalled experiences of disclosure should be sufficient to make disclosure of prognostic information our standard practice. But our findings suggest that disclosure may even have a positive effect on the parental sense of hope; physicians who provide prognostic information can support hope and trust and can help parents manage distress. Parents privately consider the best and worst outcomes for their children. Physicians who discuss these possibilities with sensitivity and realism can make the parent-physician relationship a source of hope that has the capacity to endure through many possible medical outcomes.
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
The author(s) indicated no potential conflicts of interest.
AUTHOR CONTRIBUTIONS
Conception and design: Jennifer W. Mack, Joanne Wolfe, E. Francis Cook, Holcombe E. Grier, Paul D. Cleary, Jane C. Weeks
Provision of study materials or patients: Jennifer W. Mack
Collection and assembly of data: Jennifer W. Mack, Jane C. Weeks
Data analysis and interpretation: Jennifer W. Mack, Joanne Wolfe, E. Francis Cook, Paul D. Cleary, Jane C. Weeks
Manuscript writing: Jennifer W. Mack, Jane C. Weeks
Final approval of manuscript: Jennifer W. Mack, Joanne Wolfe, E. Francis Cook, Holcombe E. Grier, Paul D. Cleary, Jane C. Weeks
Acknowledgments
We thank the parents and physicians who participated in the study, and Amy Lynch, MPH, for assistance in enrolling participants.
Footnotes
-
Supported by a fellowship from the Agency for Healthcare Research and Quality (T32 HS00063), an American Society of Clinical Oncology Young Investigator Award, and a fellowship from the Glaser Pediatric Research Network (J.W.M.).
-
Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.
- Received May 14, 2007.
- Accepted September 7, 2007.
