CONTEXT: Refusal of treatment for childhood cancer engenders much discussion. No systematic study of this phenomenon exists in countries where access to treatment is readily available.
OBJECTIVE: To identify and describe all published cases of treatment refusal for childhood cancer in the contemporary era.
DATA SOURCES: We searched PubMed, Cumulative Index to Nursing and Allied Health Literature, Scopus, LexisNexis Academic, personal database, and secondary bibliographies.
STUDY SELECTION: Eligible studies included at least 1 child <18 years of age and addressed refusal of medically recommended interventions intended to cure cancer.
DATA EXTRACTION: Cases were analyzed with respect to key features, including demographics, rationale for refusal, legal action, and medical outcome; data were combined for multiple publications discussing the same case.
RESULTS: Of 4342 unique publications identified, 579 were eligible after screening; 96 scholarly articles and 19 judicial opinions addressed 73 unique cases of treatment refusal. Most cases occurred in the United States. Rationales for refusal were broadly grouped into 4 categories. Fifty-one cases (70%) involved legal action at the time of refusal. Legal action did not reliably predict survival.
Publication bias and missing data, especially for cases without legal action, were limitations.
CONCLUSIONS: We identified important gaps in the literature, including the significant variation in approaches and lack of consensus regarding the prognostic threshold necessary for compelling treatment and the absence of voices of children and adolescents who have received treatment over their families’ objections. More research reporting effective strategies for working with families who refuse is needed.
- ALL —
- acute lymphoblastic leukemia
- HIC —
- high-income country
- LMIC —
- low- and middle-income countries
A 13-year-old boy is diagnosed with a highly curable brain tumor. With chemotherapy and radiation, he has a 95% chance of surviving 5 years without a recurrence of his disease. His parents, however, are reluctant to give permission for his treatment. Their family sees a naturopathic practitioner for routine health care and minor illnesses. They are concerned about the potential side effects and would like to try an herbal regimen first. When a postoperative scan reveals that the tumor has decreased in size dramatically, likely in response to the inflammatory stimulus of a biopsy, they are convinced that it confirms that the herbal treatment is effective. However, when he is readmitted a few months later with new seizures and the tumor has progressed, they do agree to the urgent administration of chemotherapy. After 4 cycles, with no evidence of disease on imaging, they decline to continue with radiation therapy to consolidate his remission, citing their continued concerns about late adverse effects, their religious beliefs (specifically, that their actions have no bearing on God’s will), and their accurate understanding that, in the event of relapse, cure can still be achieved with salvage chemotherapy. The pediatric oncology team discusses the case: 3 physicians are in favor of reporting the family to Child Protective Services and 2 are willing to accept the family’s decision. There is also disagreement regarding whether the patient meets criteria to be considered a “mature” minor, capable of making an autonomous decision to refuse further treatment.1 The patient’s primary oncologist elects to obtain an ethics consultation and second opinions from a pediatric radiation oncologist and a neuro-oncologist at another institution.
Refusal of treatment for potentially curable childhood cancer engenders much discussion and debate. Parents’ legal authority to make medical decisions for their children is based on 4 premises: (1) that parents generally know their children best and are, therefore, most likely to be able to make decisions consistent with their children’s best interests; (2) that they are more likely than others to be able to weigh the competing interests of other family members alongside the patient’s interests2,3; (3) that they are likely to be the most invested in the outcome of the decision; and (4) that there are clear individual and societal benefits to allowing parents to raise their children in accordance with their values, beliefs, and customs.4 This authority is tempered in situations when the parent’s decision is likely to cause substantial or irreparable harm to the child as well as in situations in which older children and adolescents, with evolving competence, are able to express disagreement.5
The diagnosis of cancer adds unique dimensions to parental decision-making. It is in 1 sense a ubiquitous diagnosis: one-quarter of American adults will die of cancer.6 This means that many parents of pediatric patients with cancer have previous experience with the disease affecting their families or close friends. It also means that for many of those parents, the experience involved death. In contrast, 80% of children diagnosed with cancer in high-income countries (HICs) are cured, and children also tolerate the adverse effects of therapy much better than older patients do.7–9 Yet the perception of chemotherapy as “poison” is pervasive around the globe and is not without some mechanistic validity.10–12 Conventional, cytotoxic chemotherapy exerts its effect on all cells in the body. Treatment-related mortality was recently estimated to be responsible for one-quarter of pediatric cancer deaths.13 The majority of children who survive cancer experience at least 1 late effect of therapy. By 20 years, 3.2% develop a second malignancy. By 30 years, almost three-quarters have a chronic medical condition, and 40% have a disabling or life-threatening condition.14–17
When parents refuse treatment of potentially curable cancer, the medical team often focuses on the certainty of death without treatment. In the background lies the smaller but still significant risk that even if treatment is eventually accepted or compelled, the child will still die of treatment-related complications or refractory disease, possibly with considerable suffering. Although many scholars have eloquently delineated the competing ethical concerns, opinions regarding how to proceed and whether to seek state intervention differ dramatically among health care providers.18–21
How and why opinions vary is poorly understood. Results from 2 recent surveys of US-based pediatricians and pediatric oncologists have indicated that considerable variation in attitudes and approaches to refusal exists.20,21 Researchers in the first study asked pediatricians and pediatric subspecialists to respond to treatment refusal in scenarios in which the patient’s age, prognosis, and parental concordance varied.20 They concluded, “When prognosis is good, best interest dominates. When prognosis is poor, parental authority is more important in younger minors, and minor autonomy is more important in older minors.” Researchers in the second study asked pediatric oncologists to respond to the statement, “As their provider I would support refusal of chemotherapy by a family,” with the following options: “Never support refusal,” “Always support refusal,” or “Support for refusal would depend on cure rate, age, or both.” Participants were then asked about 3 age and prognostic groups. Fifty-eight percent stated that their decision would depend on age and cure rate, with more than half supporting refusal for the “poor” prognosis subgroup; 11% said they would “never” support refusal.21 The social context of the decisions was not evaluated; however, evidence from other fields of medicine suggests that physicians may be more likely to seek court orders for treatment when patients belong to racial or ethnic minority groups.22
We conducted a systematic review of the published literature (including scholarly journals, legal briefs, and judicial opinions) to create a comprehensive portrait of treatment refusal for potentially curable childhood cancer in the contemporary era. We sought to describe the features of published accounts of treatment refusal for pediatric cancer, with attention to contextual features such as socioeconomic status, parental education, religion, ethnicity, and culture.
Refusal is a complex, multidimensional construct with no consensus among experts regarding terminology.23,24 The International Society of Pediatric Oncology defines “refusal” as “a decision to avoid recommended elective treatment,” separating it from “noncompliance” and “abandonment” but arguing that all 3 are consequences of a “lack of mutual understanding” between families and medical professionals.24 Abandonment is a complex phenomenon intricately related to socioecological factors, impacting individual patients and families but also occurring at an institutional level, health care system level, and societal level, and it is outside the scope of this review.23,25–27 Hinds28 has proposed 3 types of refusal by pediatric and adolescent patients: apparent, passive, and active refusal. She argues that apparent refusal is psychosocially important but not truly “refusal.” These patients say, “No, I won’t do this’ but. . .mean ‘I feel assaulted and need time to recover before I do this.” Passive refusal is defined noncompliance with treatment; patients may be quiet or polite but communicate “a desire to be [elsewhere].” Active refusal is defined as refusal rooted in conviction; patients who actively refuse are expressing, “[I am] done with this way of living.” Although we recognize that there is overlap between these categories and that research into 1 may have practical significance for the others, only Hinds’ category of active refusal was within the scope of this review.28–31
We searched 3 databases (PubMed, Scopus, and Cumulative Index to Nursing and Allied Health Literature) as well as the first author’s personal database of scholarly articles and secondary bibliographies from January 1, 1990, through December 23, 2016, for relevant articles published in the biomedical and social sciences literature. We then searched a fourth database (LexisNexis Academic) from January 1, 1990, through March 31, 2017, to include documents published in the legal literature. Where patient names were previously known or identified through full-text review as described below, those names were then used as additional search terms in LexisNexis. Search strategies tailored to each database were developed, tested, and refined in collaboration with a reference librarian (A.R.S.). The concepts of treatment refusal and cancer were defined by using combinations of controlled vocabulary (where available) and keywords. A validated search filter was used to restrict the PubMed search to the pediatric age group and was adapted for the other databases.32 Search strategies are presented in Supplemental Table 6.
Publications were eligible for inclusion if the authors (1) included at least 1 child <18 years of age or addressed issues regarding minor children generally; and (2) addressed refusal by the patient(s) and/or surrogate decision maker(s) of 1 or more medically recommended therapeutic interventions intended to cure cancer, regardless of the authors’ primary objective (Table 1). Publications in which researchers addressed refusal that did not involve cure-directed therapy were excluded. There were no exclusions for article type, language, or country of origin. Publications not available in English, French, Spanish, or Italian (languages read by the authors) were reviewed with the assistance of colleagues fluent in the relevant languages.
After duplicate elimination, all titles and abstracts were screened. The full text of eligible publications were then reviewed for inclusion. Reasons for exclusion at the full-text level were predefined and then refined during first-pass review. All full-text publications were then reviewed again and reasons for exclusion were documented (Table 2). Cases in which all events took place before 1987 were excluded, as were hypothetical or composite cases. Publications that otherwise qualified but whose authors did not address specific cases of refusal meeting these criteria were retained for future analyses. The review was designed and conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Statement.33,
All articles were reviewed by using case record forms. Data were analyzed at the case level; when multiple publications discussed the same case, data were combined. Case-level data included patient factors (eg, age, sex, mature minor designation), disease factors (eg, diagnosis, prognosis, previous therapy, response, complications), drivers of refusal (eg, concordance between parents or guardians and between parents and the patient, rationale for refusal, proposed alternatives), family and contextual factors (eg, race, ethnicity, religious beliefs, language barriers), and physician and/or institution factors (eg, size of center, years of experience, availability of treatment options such as integrative medicine) as well as the medical team’s response and the medical and legal outcomes. Rationales for refusal were coded by using an iterative process.
Patients were classified as “living” if information was available at least 1 year after refusal of treatment; otherwise, if not known to be deceased, their status was classified as “unknown.” Prognosis was defined as 5-year overall survival with standard-of-care therapy.
Whenever possible, missing data were acquired from sources outside of those identified in the systematic review. First, patient names were used as search terms in ProQuest, a database that indexes newspapers and other popular press periodicals. Sources in the popular press were then searched for specific information regarding long-term outcomes (survival) as well as family and contextual factors when that information was not available in scholarly articles and judicial opinions. Scholarly sources were searched for information regarding prognosis and standard-of-care therapy during specific treatment eras in cases where such information was not provided by articles included in the systematic review (see Supplemental References).
Risk of bias within publications was assessed qualitatively in terms of the journal’s intended audience, the publication’s objective, and the proximity of the authors to the care of the patient(s) in the case(s). For all cases with multiple sources of data, factual contradictions were also noted as a potential marker of bias.
The search identified 4342 unique publications, of which 579 were eligible after initial screening (Fig 1). 563 full-text articles were reviewed; 16 articles were unable to be obtained even after attempts to contact the authors. Of 114 scholarly articles that otherwise met eligibility criteria, 96 scholarly articles addressed specific cases of treatment refusal, as did 19 judicial opinions. Including the case reported at the beginning of this article, a total of 73 unique cases reported in the past 30 years were identified. The mean number of scholarly publications (not including judicial opinions) per case was 2.39 (median 1; range 0–21). Cases involving legal action at the time of refusal had a mean of 2.92 scholarly citations per case (median 1; range of 0–21), compared with a mean of 0.95 per case (median 1; range 0–1) for those that did not involve legal action at the time of refusal.
Demographic and Diagnostic Features
Cases are summarized in Table 3 and described individually in Supplemental Tables 7–9. Distribution of diagnoses was heterogeneous: at least 21 distinct diagnoses were represented, including leukemias, lymphomas, central nervous system tumors, and extracranial solid tumors. Nearly one-quarter of children were diagnosed with acute lymphoblastic leukemia (ALL), the most common pediatric malignancy. The majority (53 out of 73; 73%) had at least a 50% chance of 5-year survival with standard-of-care therapy, and nearly half (33 out of 73; 45%) had a good or excellent prognosis (75% or higher chance of 5-year survival with standard-of-care therapy). More than half of refusals occurred either during treatment (45%) or at the time of a relapse or diagnosis of second malignancy (8%).
Although the United States produced a plurality of reported cases (32, from at least 17 states), 7 other countries were also represented. When race was reported, most patients were white (30 out of 37); when ethnicity was reported, most patients were from the dominant or majority ethnic group within their countries of residence (25 out of 35). Notable exceptions included 3 cases involving children whose parents were immigrants or refugees, 2 cases involving First Nations children in Canada, and 4 cases involving Amish or Mennonite children in the United States. Information regarding religious affiliation was available for 30 families. Of these, 27 were reported as belonging to various Christian groups, including 13 Jehovah’s Witnesses, 3 Christian Scientists, and 3 Amish families. An additional 3 were described as religious or refusing treatment for religious reasons, without further information. Given the small number of children from racial, ethnic, or religious minority groups and the prevalence of missing demographic data, the relationship between sociocultural background and the decision to seek legal action could not be assessed.
The Context of Refusal
Discord within families regarding the decision to refuse treatment appeared to be uncommon. In 3 cases, parents openly disagreed with each other. In 1 of those 3, the 16-year-old patient’s refusal of blood was supported by her mother but not by her father; her claim to be a mature minor had been rejected by the judge. In the other 2, the patients were too young to express opinions. In an additional 4 cases, adolescent patients expressed a desire to refuse treatment that was not shared by their parents. Two of these involved patients with relapsed disease and poor prognoses choosing to forego curative treatment, and 1 involved an anxious patient who wanted to decline a biopsy. The fourth (and most prominent) involved a 14-year-old boy whose aunt and legal guardian supported his decision, as a Jehovah’s Witness, to refuse blood; his parents attempted, unsuccessfully to have him transfused over his objections.
Although in none of the publications did authors describe a situation in which parents or guardians attempted to refuse treatment that was desired by children old enough to express their wishes, judges in some cases did argue that the patients were not capable of expressing truly voluntary, informed opinions because of the influence of their parents. Of 19 cases in which the patients and families claimed that the patients were mature minors, the medical teams and/or judges agreed in 9 and rejected those claims in 10. In all the cases in which families engaged with the media, families claimed that children as young as 10 years old were in full agreement with the decision to refuse treatment.
Four major categories of rationale were identified: (1) preference for complementary and alternative medicine, (2) faith-based reasons, (3) concern about adverse effects, and (4) no insight into treatment needs (Table 4).
Responses to Refusal
In Fig 2, we depict the pathways taken in response to treatment refusal. Of 22 cases in which legal action to compel treatment was not taken, 5 of them involved patients who were living at the time of the last publication on the case, 13 involved patients who had died, and 4 involved patients who had a status that was unknown. Of 51 cases (70%) that involved legal action, 19 involved patients who were living, 16 involved patients who had died, and 16 involved patients who had an unknown status. When legal action was taken, most judges ordered treatment; only 8 ruled in favor of the family up front. Information regarding custody was often not available, but in at least 15 cases, judges declared the children involved to be dependents of the state. In Fig 3, we depict trends in the relationships between prognosis, legal action to compel treatment, and judicial rulings; however, there were examples of both patients with poor prognoses who were compelled to receive treatment and those with excellent prognoses who were not (Table 5).
The relationship between a court order and actual administration of treatment was highly variable. For example, a court order for treatment might not result in administration of treatment because of the progression of disease rendering treatment medically futile or because of a family’s decision to flee the jurisdiction. Seven decisions to compel treatment and 1 decision against compelling treatment were later reversed, sometimes after therapy had already been administered. Four were later modified, and in 2 of these, the decisions appeared to have been partly motivated by the desire to encourage a runaway teen to return home. In other cases, however, families who won the right to refuse treatment later changed their minds and voluntarily accepted therapy.
In 10 cases, criminal charges were brought against the parents after the child’s death, including 3 in which no legal action was taken before the death. In 7 cases, parents were convicted or pled guilty to lesser charges. Four cases led to proposed changes in state laws to protect parents from such charges (of which 3 were subsequently passed in state legislatures). In 4 cases, parents were documented as having brought countersuits against the hospital, state child protection agency, or affiliated individuals.
Risk of Bias
It was not possible to assess the quality of case information for cases cited by only a single article. Of the 23 cases (32%) discussed by more than 1 source (involving 68 scholarly articles), only 2 instances of factual contradictions regarding details of a specific case were observed (once regarding the prognosis, which was verified with an outside source; once regarding a family’s rationale for refusal). Target audience and objective may have influenced the type of information included but did not appear to affect the veracity of the information provided.
Across all publications, narrative details regarding the medical team’s approach to the refusal were seldom available. Ethics consultations were clearly documented as having been obtained in only 5 cases, whereas the offer of a compromise between the family and the medical team was documented in 11 cases. Physician and institution factors (such as the treating physician’s previous experience or the availability of integrative medicine at the treating institution) were never reported.
This systematic review is larger by several orders of magnitude than previously published case series and narrative reviews describing treatment refusal for potentially curable childhood cancer in countries where access to such treatment is readily available.18,34 Our findings support the conclusion that there is a remarkable mismatch between the scholarly attention paid to the subject of treatment refusal in this setting and what is actually known about refusal.
Although we identified many gaps in available data, we also observed important trends. The decision to seek legal action was not consistently predictive of outcome. This is likely because legal action did not consistently result in treatment being received. When treatment was received, it was often administered after significant delays because of the nature of the legal process. Summarizing several cases, legal scholar Derry Ridgway noted that “adversarial” questioning in the courtroom failed to uncover nuances of each child’s medical situation that might have led medical providers to propose and quickly implement acceptable compromises.35 In addition, although it cannot be proven from this data, the stress of legal action, sometimes including involvement of Child Protective Services, family separation, and foster care, almost certainly adversely impacted children’s psychosocial well-being and perhaps physical health.36,37
Crucial questions remained unanswered. Information regarding the medical team’s initial response to the refusal (presuming that most providers attempt at least some dialogue with the family before seeking a court order) was lacking from most publications. Equally few provided complete information regarding the socioeconomic, cultural, and religious background of the patients and families, despite a recent global survey of providers’ experiences with refusal that indicated that providers in HICs were more likely than those in low- and middle-income countries (LMIC) to regard religious faith as predictive of refusal. The same survey participants expressed concern that immigration status and language barriers might impact rates of treatment refusal.25 Because we were unable to identify race or ethnicity for almost half of the included patients, we were unable to even hypothesize about the relationship between these factors and providers’ responses to refusal. These are critical gaps in our understanding of the context of treatment refusal.
Studying treatment refusal is understandably challenging: cases occur infrequently and unpredictably in HICs.26,27 Patients whose parents refuse to grant permission for treatment are unlikely to consent to participation in clinical research studies that systematically track outcomes. No center has published a study or proposed a method of tracking the decision-making process prospectively. Retrospective studies are likely to suffer from recall bias, identifying only the most controversial and high-profile cases: those involving protracted legal battles and media attention. Cases in which health care providers and families successfully negotiate their differences and reach a compromise may go unnoticed and unreported.34,38 However, the fact that almost half of families first refused treatment during therapy suggests that there are opportunities, likely beginning at the time of diagnosis, to identify families at risk for interruption of treatment and to intervene.
Although researchers in many studies of shared decision-making in pediatric oncology have engaged patients and parents, only a single article written by a family member was identified in this search, and in only 1 article included in the review did researchers suggest that the family had approved the discussion of their child’s case.39–43 For this reason, the process of coding rationale for refusal raised conceptual questions and proved to be a primarily exploratory and hypothesis-generating endeavor. A proposed conceptual map for how the categories of refusal might intersect and interact is presented in Fig 4.
Impressively, in a couple of studies of treatment abandonment in LMIC, researchers have achieved precisely this goal of appreciating family perspectives.44,45 Rossell et al45 conducted in-depth, semistructured interviews with 41 caregivers of children with cancer in El Salvador, all of whom were identified as being at-risk of abandoning treatment or who had already abandoned treatment. They sought to understand how caregivers’ beliefs in miraculous cures might influence their decisions. Similar studies, perhaps building on ethnographic work exploring the experiences of pediatric nurses and adapted for the unique cultural and structural context of each country, would be useful to pediatric health care professionals facing treatment refusal in high-income settings.46
Like all systematic reviews, publication bias is a major limitation of our findings. There is a clear propensity toward publication of cases in which legal action was pursued to compel treatment; this was, of course, exacerbated by our inclusion of a legal database and our deliberate search for relevant judicial opinions. However, this bias was observed even in the academic literature, perhaps in part because physicians caring directly for these patients may be reluctant to publish those cases, and thus scholars interested in the topic tend to analyze cases that have entered the public domain. Regardless, the unfortunate consequence is that cases in which medical teams reached a successful compromise with a patient or family who initially refused are underrepresented, although they may be the most valuable in terms of practical guidance for providers.
Secondly, missing data may have affected our analysis. Many of articles were written by authors who were themselves not directly involved in the care of the patients and families and/or were written in support of a specific ethical or legal argument. More complete information (from multiple sources) was available exclusively for cases that involved legal action; these were typically the same cases in which patients’ real names had entered the public domain, allowing us to use popular press sources to complete the data set. These limitations mean that authors (1) may not have had access to all the critical information regarding a case or may have had incorrect information and/or (2) may have consciously or unconsciously presented the case in the way that best supported their arguments.
Finally, although we chose to precisely define the scope of this review, there is likely overlap between different types of refusal in practice. We excluded articles that discussed refusal of non–cure-directed interventions, including supportive care measures, because refusal of these interventions has generally been viewed as within the scope of parental authority. However, it is worth noting that persistent and repetitive refusal of or nonadherence to adjunctive therapies can sometimes compromise patients’ long-term outcomes. Both ethically and medically, the dichotomy between these may be artificial, better suited to classrooms than hospital rooms.
This review highlighted notable gaps in the literature, including the significant variation in approaches and lack of consensus regarding the prognostic threshold necessary for compelling treatment, the absence of voices of children and adolescents who have received treatment over their families’ objections, and the lack of any systematic attempts to study or report effective strategies for working with families who refuse curative treatment. That we failed to find an association between the decision to seek legal action and positive outcomes in terms of survival likely reflects the nature of the legal process itself but also supports the need for further research to achieve optimal outcomes.
Dr Caruso Brown is grateful to John Lantos, MD, and the faculty and staff of the Bioethics Center at Children’s Mercy Hospital, Kansas City, Missouri, for their support of this project; Alexandra Kolenová, MD, PhD, Chief of Pediatric Oncology at Comenius University in Bratislava, Slovakia, for providing invaluable opportunities to speak about this topic with clinicians in Central Europe during the project’s conception; and Kathy Faber-Langendoen, MD, Chair of the Center for Bioethics and Humanities at State University of New York Upstate Medical University for her support and guidance. The authors also thank the reference staff at the Syracuse University College of Law Library for their guidance with legal literature, the staff of the Document Delivery department at Upstate Medical University’s Health Sciences Library for their assistance with acquiring full-text articles, and Elizabeth Oaks Krieger, MD, for her assistance with data collection.
- Accepted August 29, 2017.
- Address correspondence to Amy E. Caruso Brown, MD, MSc, MSCS, Center for Bioethics and Humanities, 618 Irving Ave, SUNY Upstate Medical University, Syracuse, NY 13210. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Hurria A,
- Wildes T,
- Blair SL, et al
- Mukherjee S.
- Turcotte LM,
- Whitton JA,
- Friedman DL, et al
- Hord JD,
- Rehman W,
- Hannon P,
- Anderson-Shaw L,
- Schmidt ML
- Mercurio MR
- Spinetta JJ,
- Masera G,
- Eden T, et al
- Itriago E,
- Power-Hays A,
- Rodriguez-Galindo C,
- Friedrich P
- Robertson EG,
- Wakefield CE,
- Marshall KH,
- Sansom-Daly UM
- Ridgway D
- Edwards VJ,
- Anda RF,
- Dube SR,
- Dong M,
- Chapman DF,
- Felitti VJ
- Friebert S,
- Kodish E
- Copyright © 2017 by the American Academy of Pediatrics