Adolescent and Young Adult Cancer Patients’ Experiences With Treatment Decision-making

We surveyed patients with cancer aged 15 to 29 years at diagnosis and their oncologists at Dana-Farber Cancer Institute in Boston, Massachusetts. Baseline surveys were administered from April 2014 to October 2017, with follow-up surveys completed in November 2018.⁹ English-speaking patients were eligible if they were 1 to 6 weeks from diagnosis at first contact and oncologists provided permission for contact. Oncologists were asked to complete a questionnaire on patient prognosis when permission was provided.

Eligible patients were given a letter inviting participation, the survey, and an opt-out postcard. Whenever possible, patients were approached in person (n = 199); some were approached by mail (n = 76). For patients <18 years old, patient assent and parent- and/or guardian-signed consent was required; for patients ≥18 years old, documentation of informed consent was waived. Surveys were offered in paper or electronic format; electronic versions were completed in person via tablet or remotely via a secure link. Nonrespondents received up to 2 subsequent contacts. Return of the baseline questionnaire was required within 12 weeks of diagnosis for inclusion. Patients received a $50 gift card for participation. Living patients who completed the baseline questionnaire were offered follow-up questionnaires 4 and 12 months after diagnosis.

Of 303 eligible patients, 275 were approached, 218 consented, and 203 completed the baseline survey (74% of eligible and approached; 67% of all eligible). A total of 99 oncologists completed the item on prognosis for 185 of 203 (91%) participants. A total of 177 patients completed 4-month surveys, and 154 completed all 3 assessments (51% of 303 eligible at baseline and 76% of 203 baseline survey participants; 10 died between baseline and 12-month surveys).

Patient questionnaires included previously developed items^10,11 and limited new questions developed by using patient interviews, previous work, and literature review. Conceptual, face, and content validity were established through pilot testing with 11 AYA patients across the age range. Surveys required 30 to 40 minutes to complete.

Surveys focused on patients’ experiences with cancer-related communication and decision-making. In the baseline survey, patients were asked to “think about the time when you were first diagnosed with cancer and decisions were made about the treatment you would receive.” Patient decision-making preferences and roles were assessed by using questions¹² developed by Degner and Sloan,¹³ including “the role you would prefer” and “the role you actually played” when making decisions about cancer treatment. Separate questions asked about roles relative to the oncologist, parent and/or guardian if applicable, and spouse and/or partner if applicable. Response categories were “I prefer to make decisions about treatment with little or no input from my oncologist (or parent or spouse and/or partner),” “I prefer to make decisions after considering my oncologist’s opinion,” “I prefer that my oncologist and I make decisions together,” “I prefer that my oncologist make decisions after considering my opinion,” and “I prefer that my oncologist make decisions with little or no input from me.” Patients were asked how oncologists discussed treatment options (“Oncologists told me what my treatment would be, without offering other choices,” “Oncologists offered different choices for treatment but told me 1 particular choice was the best 1 for me,” and “Oncologists offered different choices for treatment and allowed me to decide which was best for me”).

Decisional regret was assessed by using the Decisional Regret Scale¹⁴ administered at all 3 time points. Patients responded to the following 5 statements: “I have made the right decisions,” “I regret the choices that were made,” “I would make the same choices if I had to do it all over again,” “the choices caused harm,” and “the decisions were wise.” Response options were “strongly agree,” “agree,” “neither agree nor disagree,” “disagree,” or “strongly disagree.” The 16-item Decisional Conflict Scale was used to measure patient uncertainty about decisions as well as feeling uninformed, unclear about personal values, and unsupported in decision-making.¹⁵ 

Additional questions assessed factors hypothesized to be associated with regret (detailed below).

Patients were asked to report sex, race and/or ethnicity, educational attainment, precancer living situation, level of financial dependence on parent(s) and/or guardian(s) before diagnosis, and current living situation. Age and diagnosis were determined by using medical records. Patient prognosis was evaluated by using oncologist report; oncologists were asked, “How likely do you think it is that this patient will be cured of cancer?” with response categories of “extremely likely" (>90% chance of cure), “very likely" (75%–90%), “moderately likely" (50%–74%), “somewhat likely" (25%–49%), “unlikely" (10%–24%), “very unlikely" (<10%), or “no chance of cure.”^11,16,17 

We assessed communication quality^18,19 with questions on physician sensitivity (“How often does your oncologist convey information in a sensitive manner?”), time for questions (“How often does your oncologist take enough time to answer your questions?”), and clarity (“When you ask questions, how often do you get answers that are understandable?”). Response categories were “always,” “usually,” “sometimes,” or “never.” Quality of information^11,20 was assessed about diagnosis, treatment, prognosis, long-term effects, cause of cancer, and treatment response (“excellent,” “very good,” “good,” “fair,” or “poor”). Prognostic disclosure was evaluated by using 5 items^20,21 assessing whether prognosis was discussed, the oncologist offered information or the patient asked, prognosis was discussed as a number, written information was provided, and the patient wanted additional information. The questionnaire included 1 item from the Trust in Physicians Scale (“How much do you trust your oncologist’s judgment about your medical care?”).^22,23 Patients were also asked, “How well do you think your oncologist understood what was most important to you when your treatment started?” (“extremely well,” “very well,” “somewhat well,” “a little,” or “not at all”).

The Hospital Anxiety and Depression Scale (HADS), administered at all 3 time points, evaluated depression and anxiety.²⁴ Peace of mind was assessed by using 5 items from the Functional Assessment of Chronic Illness Therapy–Spiritual Well-being Scale.²⁵ 

The institutional review board of Dana-Farber Cancer Institute approved this study.

Measures were dichotomized for analysis, consistent with previous work, with cutoffs specified a priori. Decision-making preferences and roles with oncologists were evaluated in 3 categories: patient led (if patients made decisions after considering oncologists’ opinions or with little or no input from oncologists), shared (if patients made decisions together with oncologists), or oncologist led (if oncologists made decisions after considering patients’ opinions or with little or no input from patients). Analogous categories were used with respect to parents and partners. Holding one’s ideal role in decision-making was defined as a match between patients’ preferred and actual roles. Information and communication quality were summed and dichotomized at the median for analysis.^10,26,27 Prognostic disclosure was dichotomized as 0 to 2 versus >2 elements.²⁰ Anxiety and depression were defined as scores ≥8 on the HADS subscales.²⁴ The χ² tests evaluated differences in proportions of categorical variables; the McNemar-Bowker test evaluated symmetry between preferred and actual decisional roles in the 3-category variable construction to determine if disagreement in preferred and actual roles was predominantly in 1 direction (eg, over- or under-engagement) or evenly distributed.

We analyzed the Decisional Regret Scale using previously reported methods.^14,28,29 We obtained mean scores from 1 to 5, with reverse scoring when appropriate, in which 1 indicated the least regret and 5 the most regret. Scores were decreased by 1 and multiplied by 25 for a range of 0 to 100. As previously specified,^28,29 scores of >25 were categorized as heightened regret, which served as a dichotomous outcome for logistic regression. A backward selection procedure was used to construct multivariable models; the significance level for removal was 0.05. Patient age was included in multivariable models regardless of significance.

The sample size was designed to offer 80% power to detect a 20% difference in information (desire for detail about prognosis reported separately⁹) and decision-making needs (desire to hold primary role in decision-making relative to oncologists and to parents) by age at 2-sided α = .05. Statistical analyses were conducted by using SAS version 9.4 (SAS Institute, Inc, Cary, NC).

One-quarter of participants were aged 15 to 17 years (25%), 21% were 18 to 21 years, and more than half were 22 to 29 years (54%; Table 1). Lymphoma was the most common diagnosis (32%), followed by sarcoma (16%), genitourinary cancer (14%), and leukemia (13%). Most patients were white and non-Hispanic.

More than half of patients (58%; Table 2) preferred to make decisions together with oncologists; 20% wanted to hold primary responsibility for decision-making, and 22% wanted the oncologist to hold primary responsibility. Preferences did not differ statistically by age (P = .07), although the absolute proportion of the youngest AYAs wanting shared decision-making was lower than among older patients (41% of 15–17-year-olds; 67% of 18–21-year-olds; 62% of 22–29-year-olds). When asked about the roles they actually held, 66% (131 of 199) held their ideal roles relative to oncologists, with 22% (43 of 199) holding more passive roles than desired and 13% (25 of 199) holding more active roles than desired (test of symmetry P < .0001). The proportion of patients who held their ideal roles did not differ by age (P = .99).

Among patients with a living parent and/or guardian (n = 197), 51% wanted to make decisions after considering the parent’s opinion, and 39% wanted to make decisions together; only 10% wanted parents to lead decision-making. A total of 81% of patients were able to hold preferred roles relative to parents. Preferred and actual roles differed by age (P < .0001 for both), with younger patients wanting and receiving more parental involvement than older patients. However, younger patients were also more likely to be less involved than they wanted relative to parents (29% among 15–17-year-olds, 12% among 18–21-year-olds, and 6% among >21-year-olds; P = .002). Among patients with an involved partner or spouse (n = 95), 41% preferred shared decision-making, and 85% held their ideal roles.

When asked how oncologists discussed treatment choices at diagnosis, 36% (71 of 196) reported that oncologists told them what treatment would be without offering other choices; 46% (90 of 196) reported that oncologists offered different choices for treatment but said 1 particular choice was best, and 18% (35 of 196) reported that oncologists offered different choices and allowed the patient to decide which was best. Younger patients were more likely to report that the oncologist told them what treatment would be without offering choices (51% of 15–17-year-olds, 31% of 18–21-year-olds, and 31% of >21-year-olds; P = .05).

Nearly one-quarter of patients (24%; 47 of 195) had decisional regret scores indicative of heightened regret about initial cancer treatment decisions. Regret was associated with scores on the Decisional Conflict Scale, including the total score (P < .0001; Table 3) and all subscales; patients with regret had higher scores for uncertainty about the best choice, feeling uninformed, feeling unclear about personal values, and feeling unsupported in decision-making (P < .0001 for all subscales), indicating challenges across multiple domains.

Bivariable analyses between regret and patient and communication attributes are shown in Table 4. In a multivariable model adjusted for age, regret was less likely among patients who trusted the oncologist completely (odds ratio [OR] 0.17 [95% confidence interval (CI) 0.06–0.46]; P < .001; Table 5) and who reported that the oncologist understood what was important to them when treatment started (OR 0.13 [95% CI 0.04–0.42]; P < .001). Regret was also associated with prognosis (P = .04). Although Hispanic patients had higher odds of regret relative to white patients, regret was not statistically associated with race and/or ethnicity across all categories (P = .13).

Aggregate rates of regret remained similar 4 and 12 months postdiagnosis, with 23% evidencing regret at 4 months (P = .27 relative to baseline) and 19% evidencing regret at 12 months (P = .85 relative to baseline). A total of 49 of 154 patients (32%) reported having made a new cancer treatment decision by 12 months, and 25% had heightened regret about the initial decision. Six patients relapsed by the 4-month survey; 50% experienced regret at 4 months relative to 23% of nonrelapsed patients (P = .14; χ² test). By 12 months, 11 patients had relapsed; 45% experienced regret at 12 months relative to 18% of nonrelapsed patients (P = .03).

Baseline regret was associated with increased anxiety (54% among those with regret versus 29% among those without; P = .01; χ² test) and depression (39% vs 20%; P = .03) at 4 months but not at 12 months (33% vs 33% [P = .97] for anxiety and 23% vs 13% [P = .18] for depression). Regret at 12 months was associated with increased anxiety (59% vs 26%; P = .001) and depression (28% vs 12%; P = .03) at 12 months.

Although researchers have previously raised concerns about AYA patients’ involvement in decisions about cancer care,^2,5,6 our study found that most AYAs were highly engaged in decision-making. Most wanted to either make decisions together with oncologists, parents, or partners or hold primary responsibility for decisions themselves, and most held the roles they wanted in decisions about care.

Of note, the youngest patients in our study were less likely to hold the roles they wished in decision-making. Although many wanted parental involvement, parents tended to take on more decisional involvement than patients wished. Clinicians who care for adolescents may need to help navigate complex needs for support and autonomy, offering parental engagement but ensuring that adolescents have the opportunity to express their wishes about care.

Nearly one-quarter of AYAs experienced regret about their initial treatment decision, a proportion that remained stable over the first year after diagnosis. Regret at diagnosis was associated with higher rates of depression and anxiety 4 months later, suggesting corresponding psychological distress. Although we did not ask the AYAs in our study about the source of regret, the concept of regret is often linked to poor outcomes¹⁴ (as among adults with breast or prostate cancer who have negative outcomes after local control).^29,30 We did find that regret was higher among the small number of patients who relapsed by 12 months from diagnosis, but most patients with regret had not experienced recurrence, and this finding does not explain the prevalence of regret soon after diagnosis. Our findings therefore raise the possibility that regret is at times more reflective of the communication process than medical outcomes, suggesting areas for potential improvement among clinicians who care for these patients.

Although we had hypothesized that regret would be more common among patients with lower decisional engagement, decisional roles were less important than AYA patients’ relationships with oncologists, especially the extent to which patients trusted oncologists and felt that oncologists understood their personal values. The National Academy of Medicine defines “continuous healing relationships” as a central aspect of the patient experience³¹; patients want to be listened to and respected, to hold relationships of trust, and to be known and understood as individuals.³² These attributes may be especially important to AYAs, who are forming identities, moving toward independence, and negotiating new kinds of relationships with adults. Researchers have previously shown that AYAs are highly connected to clinicians, value a strong therapeutic alliance, and look to clinicians for presence and support throughout care.^33,–36 Our results highlight the importance of oncologists taking the time to ask patients about their lives, hopes and fears, and priorities as they work together to make plans for treatment and further solidify the oncologist-patient relationship as a core component of decision-making for AYAs.

Although a positive relationship with the oncologist appears to be a critical element of decision-making for AYAs, patients also need to be well informed about treatment options, alternatives, and risks and benefits, and they need the opportunity to weigh options in the context of their lives and priorities. In addition, ambivalence about decisions, a normal phenomenon throughout life, may be elevated among young people.^37,38 AYAs may be less accustomed to uncertainty than older adults, have less developed coping skills for dealing with complex decisions,^39,40 and are often still developing clarity around personal values. They may also be in conflict around issues of interdependence with family members whose involvement in decision-making may complicate AYAs’ assessments of treatment decisions. Yet most AYAs in this study were able to overcome these conflicts to reach greater satisfaction with decisions.

This work is limited by the study of a small population of AYAs at a single large academic cancer center with limited diversity. AYAs who seek care at academic centers may differ from those whose care is at community sites. We have identified associations between clinical and communication factors and regret, but cannot presume that these factors are causative; in fact, regret may have caused patients to reflect differently on communication experiences rather than the reverse. Regret may also reflect psychological vulnerabilities of patients rather than their clinical experiences. We have performed multiple statistical analyses, which could have led to false-positive findings. Communication interactions were not directly observed; survey responses therefore reflect patients’ perceptions rather than objective descriptions of what occurred. Finally, although we focused on regret associated with initial treatment decisions, patients may continue to reflect and reconsider decisions throughout care. Although our longitudinal data provide some insight, these issues deserve attention in future work.

Our findings raise intriguing questions about decisional regret among AYAs, opening a new area for future research. Future work should delve more deeply into the reasons for regret among AYAs and, ultimately, consider interventions to optimally support AYA decision-making and reduce negative psychosocial outcomes. In addition, our results suggest that relationships of trust and mutual understanding may serve as a foundation for effective cancer treatment decision-making among AYAs. Oncologists should approach decision-making conversations with AYAs with a goal of understanding what is important to the young patient in the room. Understanding patient values, preferences, and priorities is essential to formulating an optimal treatment plan for AYAs and all patients.

AYA

adolescent and young adult

CI

confidence interval

HADS

Hospital Anxiety and Depression Scale

OR

odds ratio