Longitudinal Prognostic Communication Needs of Adolescents and Young Adults with Cancer

Abstract

BACKGROUND –

Although most adolescent and young adult (AYA) patients with cancer desire prognostic information, little is known regarding how preferences for prognostic communication change over time.

METHODS –

Longitudinal, prospective questionnaire-based cohort study of 136 AYA cancer patients aged 15-29, treated at a large academic cancer center. Previously published scales were administered at time of diagnosis, and 4 and 12 months after diagnosis.

RESULTS –

Most patients reported that prognostic information was very/extremely important at diagnosis (85%), 4 months (96%, p=.002 relative to baseline), and 12 months (81%, p=.02 relative to baseline). Few patients reported that prognostic knowledge was very/extremely upsetting at baseline (7%), 4 months (9%, p=.44 relative to baseline), or 12 months (11%, p=.27 relative to baseline). Most patients were satisfied with the amount of prognostic information received throughout the year after diagnosis (81%, 86%, and 81% at diagnosis, 4 months, and 12 months, respectively). This proportion did not change between diagnosis and 4 months (p=.16) or between diagnosis and 12 months (p=1.00). In multivariable analysis, satisfaction with prognostic information was associated with patient report of high-quality communication (OR 2.67, 95%CI 1.38 to 5.17) and >75% chance of cure (OR 2.39, 95%CI 1.24 to 4.61) after adjustment for patient age category, race/ethnicity, and time point of administration.

CONCLUSION –

Most AYA patients were satisfied with prognostic disclosure over time, but a sizeable minority wanted additional information. Desire for prognostic information increased over time. Clinicians should return to prognostic discussions over time to support AYA patients with cancer.

Precis:

Most adolescents and young adults (AYAs) with cancer highly valued prognostic information throughout the year after diagnosis. While many AYAs were satisfied with prognostic information, a sizeable minority had unmet needs over time.

Introduction

Prognostic disclosure in oncology is a challenging yet essential task. In adult oncology, receiving prognostic information helps patients to prepare for the future, especially when prognosis is poor.^{1, 2} However, awareness of poor prognosis for some patients has also been associated with worse quality of life, especially for patients who lack active coping strategies.³ In pediatric oncology, parents consistently report a desire for prognostic information even when cure is unlikely,⁴ and prognostic communication has been associated with improved hope,⁵ trust in the oncologist,⁶ satisfaction with medical care,⁷ and peace of mind.⁸ This desire for prognostic information persists over time, regardless of prognosis.⁹ Given the emotional burden of a poor prognosis, disclosure has perpetually challenged clinicians. Perhaps as a result, many adult patients and parents of children with cancer overestimate the prospect of cure.^{9, 10}

Few studies have specifically evaluated prognostic communication for adolescents and young adults (AYAs) with cancer, and none have evaluated this communication longitudinally. The AYA period is a time of marked transitions, including physical, developmental, and social changes, and the broad psychosocial concerns of AYA patients with cancer are often insufficiently addressed by clinicians.¹¹ From a few small studies, we know that AYA patients desire communication tailored to their age and development.¹² Additionally, we know that 10-37% of AYA patients report unmet information needs.¹³ Furthermore, qualitative studies suggest that many AYA patients misestimate prognosis,¹⁴ and some patients feel harmed by prognostic disclosure from oncologists who are aloof, patronizing, or cold.¹⁵

AYA patients with cancer face a life-threatening diagnosis at a young age. At this stage of development, younger AYAs often live in the present without full considerations about future consequences, yet they progress toward a time when they will plan for the future with a sense of agency. As such, it is uncertain whether AYA patients’ desire for prognostic information would change over time. Our group previously reported baseline data on a cohort study of AYA patients with cancer that explored prognostic awareness and outcomes of disclosure.¹⁶ This was the largest study to date analyzing prognostic communication with AYA patients. In this initial analysis, 83% of patients considered prognostic information very/extremely important, and more extensive prognostic disclosure was associated with higher trust in the oncologist, peace of mind, and hope. Greater disclosure was also associated with lower prognosis-related distress. Herein, we present a longitudinal analysis of this cohort with repeated evaluations at 4 and 12 months. Our primary purpose was to determine whether desire for prognostic information and satisfaction with information received changed over time. Secondarily, we evaluated for predictors of longitudinal satisfaction with prognostic disclosure. We hypothesized that desire for prognostic information would remain stable, but satisfaction with disclosure would decrease.

Methods

As previously described, we surveyed AYA patients with cancer between 15 and 29 years of age at diagnosis and their oncologists at Dana-Farber Cancer Institute (DFCI; Boston, MA) between April 2014 and October 2017.¹⁶ Although AYA age range in the US is 15-39, we studied a narrower spectrum to focus on communication issues that younger patients might experience. Patients were eligible if they spoke and read English, were 1-6 weeks from diagnosis at first contact, and their oncologist permitted contact. Eligible patients were identified through review of clinic and medical records.

303 patients were eligible during the study time period, and 275 were approached (28 missed due to logistical challenges). Patients younger than 18 years provided assent and parents/guardians provided consent. Documentation of informed consent was waived per DFCI institutional review board for patients 18 years or older. Surveys were offered in paper or electronic format. To be included, patients had to return the questionnaire within 12 weeks of diagnosis.

Four months and 12 months after diagnosis, we contacted patients who had previously completed questionnaires, requesting that they complete follow-up questionnaires. We offered participants $50 gift cards upon completion.

In total, 154 patients completed this questionnaire. (Figure 1) Of these 154 patients, 18 were excluded from the current analysis because of missing values for the primary outcome (satisfaction with prognostic disclosure) at any time point, leaving a total analytic cohort of 136 patients.

Figure 1.

Enrollment Flow Chart

Data Collection

Survey development has been described previously.¹⁶ Patient questionnaires included items from previously developed surveys in medical and pediatric oncology,^{17, 18} and select new items developed based on patient interviews, review of literature, and prior work. We established conceptual, face, and content validity through pilot testing with 11 AYA patients.¹⁶ The survey assessed patients’ needs and experiences related to communication about prognosis, which we defined as “the chances that your cancer would be cured.” Oncologists completed a question on patient prognosis when permission was provided.

Patients reported their gender, race or ethnicity, and educational attainment on the baseline questionnaire. Age and diagnosis were determined from medical records. This study was approved by the DFCI Institutional Review Board.

Prognosis Communication Preferences and Experiences

We assessed desire for prognostic information by asking patients how important it was “to know about your likelihood of being cured,” “to have a precise understanding of your chances of being cured,” and “how upsetting” it was to know the likelihood of cure. Each question contained 5 response options: “extremely,” “very,” “somewhat,” “a little,” or “not at all important.”^{17, 18}

The primary outcome for this analysis was patient satisfaction with the amount of prognostic information they received, which we assessed using a previously published scale.^{17, 19} We asked patients, “All things considered, how do you feel about the amount of information you know about your likelihood of being cured?” with 3 response options: 1) I wish I had more information about my likelihood of cure; 2) I now have about the right amount of information; 3) I wish I had less information about my likelihood of cure.

Communication Characteristics

We assessed patient perspectives on the quality of information received about diagnosis, treatment, prognosis, long-term effects, cause of cancer and treatment response, with responses of “excellent,” very good,” “good,” “fair,” or “poor.”¹⁸ We assessed patient perceptions of the quality of the physician communication style with previously published items that included questions about physician sensitivity, time allowed for questions, and clarity of answers. Response options were “always,” “usually,” “sometimes,” and “never.”^{20, 21}

Psychosocial Factors

We evaluated patient peace of mind with 5 items from the Functional Assessment of Chronic Illness Therapy – Spiritual Well-Being Scale.²² We evaluated trust in the oncologist using an item from the Trust in Physician Scale (“How much do you trust your oncologist’s judgment about your medical care?”) with response options of “completely,” “mostly,” “somewhat,” “a little,” or “not at all.”²³ We evaluated anxiety and depression with the Hospital Anxiety and Depression Scale.²⁴ Decisional regret was evaluated by the Decision Regret Scale.²⁵ We also included a new item based on pilot interviews: “How much do you like your oncologist?” (“a great deal,” “quite a bit,” “somewhat,” or “not at all”). See Supplementary Table S1 for full descriptions of measures.

Statistical Analysis

Measures were dichotomized consistent with previous work to allow for comparison with past studies.^{5, 17, 18, 22} For responses to questions about desire for prognostic information and desire for numeric information, we categorized responses as very/extremely important versus all other responses. Communication and information quality were dichotomized at the median.¹⁷ Trust was dichotomized as a response of completely versus any other response in keeping with prior work.⁵ This dichotomization of trust is consistent with evidence that a single negative communication encounter can have lasting effects.^{22, 26} For the Hospital Anxiety and Depression subscales, a score of greater than 7 is the accepted cutoff.²⁴

We used McNemar’s test to evaluate for marginal symmetry at baseline, 4-month, and 12-month time points for the participant reports of the following outcomes: preferences for prognostic information, whether prognostic information was very/extremely upsetting, and satisfaction with amount of prognostic information. Subsequently, we performed exploratory subgroup analyses based on age range, also using McNemar’s test, albeit with limited power due to small subgroups. Given that only 1 participant reported a preference for less prognostic information at baseline and 4 months, these patients were excluded from this analysis for marginal symmetry. Additionally, we evaluated for association between being very/extremely upset by prognostic information and having >75% chance of cure at each time point using Fisher’s Exact test due to cell counts <5. We also evaluated for association of being upset by prognostic information with desiring more information at each time point using Chi-Square test.

We then evaluated for factors associated longitudinally with “having the right amount of prognostic information” using logistic regression. For bivariable analyses, we used logistic regression with random effects modeling for time in order to isolate effects of the independent variables. Some independent variables were considered at a single, baseline time point (race/ethnicity, education, gender, physician-rated prognosis at baseline, and participant report of high-quality information at baseline), while others were time varying covariates with longitudinal responses (complete trust in physician, liking oncologist a great deal, relapsed disease status at 4 and 12 months, and participant report of high-quality communication).

For multivariable logistic regression, we employed generalized estimating equations adjusted for the time point of survey administration. We utilized a backward elimination technique to create a multivariable model of factors associated with “having the right amount of prognostic information,” with criteria for entry of p=0.10 and retention of p=0.05. Race/ethnicity, age, prognosis, and time point were included regardless of significance. Finally, we performed exploratory bivariable logistic regression with psychosocial outcomes (anxiety, depression, regret, and peace of mind) as the dependent variables and “having the right amount of prognostic information” as the independent variable. These bivariable analyses utilized mixed linear models that employed random effects modeling for time. Statistical analyses were conducted using SAS statistical package v9.4 (SAS Institute, Cary, NC).

Results

Participants were predominantly white and non-Hispanic. Lymphoma (32%) was the most common malignancy, with similar proportions of genitourinary cancers (15%), sarcomas (13%), leukemia (13%), and breast cancer (11%). (Table 1)

Table 1.

Patient characteristics.

Age, years	N (%)
15-17	28 (21)
18-21	28 (21)
22-29	80 (58)
Gender
Female	67 (49)
Ethnicity
Non-Hispanic	127 (93)
Hispanic	9 (7)
Race/ethnicity*
White	117 (87)
Black or African American	6 (4)
Asian or Pacific Islander	8 (6)
Native American or Other	5 (3)
Cancer Diagnosis
Leukemia	18 (13)
Lymphoma	44 (32)
Breast	15 (11)
Genitourinary	20 (15)
Sarcoma	17 (13)
Other Solid Tumor	16 (12)
Brain	6 (4)
Educational attainment
Some high school	28 (21)
High school graduate or equivalent	18 (13)
Some college/technical school	23 (17)
College graduate	50 (37)
Graduate/professional school	17 (12)
Patient has children	10 (7)
Oncologist-rated prognosis (chance of cure), %
Extremely likely (> 90)	30 (25)
Very likely (75-90)	51 (42)
Moderately likely (50-74)	19 (16)
Somewhat likely (25-49)	10 (9)
Unlikely (10-24)	4 (3)
Very unlikely (< 10)	4 (3)
No chance of cure	3 (2)

n=136 with complete prognostic satisfaction data at all 3 time points. Missing data: race (n=2); has children (n=1); prognosis (n=15).

^*Not mutually exclusive

Characteristics of Prognostic Communication over Time

Most patients considered knowing about chances of cure to be very/extremely important throughout the first year after diagnosis, and this proportion increased over time: at baseline (85%, 116/136), 4 months (96%, 130/136, p=.002 relative to baseline), and 12 months (93%, 126/136, p=.02 relative to baseline). We present exploratory analyses of these preferences by age group in Figure 2. Similarly, most patients reported that having a precise, numeric understanding of prognosis was very/extremely important at baseline (76%, 103/136); this proportion increased at 4 months (85%, 116/136, p=.01 relative to baseline) and 12 months (85%, 115/136, p=.01 relative to baseline). Few patients reported that knowing about prognosis was very/extremely upsetting at baseline (7%, 9/136), 4 months (9%, 12/136, p=.44 relative to baseline), or 12 months (11%, 15/136, p=.27 relative to baseline). Being upset by prognostic information was not associated with worse prognosis (<75% chance of cure) at any time point (p=1.0 at baseline, p=.30 at 4 months, and p=1.0 at 12 months), but was associated with desiring more prognostic information at 4 months (p=.003 by Chi-square testing) and 12 months (p=.03), without reaching significance at diagnosis (p=.07).

Figure 2. Percentage of Patients Reporting Prognostic Information Was Very or Extremely Important.

Total n=136. 15-17 years, n=28; 18-21 years, n=28; 22 to 29 years, n=80. * P<.05 by McNemar’s test of symmetry, relative to baseline.

At diagnosis, 81% (108/134) of patients reported satisfaction with the amount of prognostic information received, compared to 86% (115/134) at 4 months, and 81% (108/134) at 12 months. Only 1 participant at diagnosis and 1 participant at 4 months reported a desire for less prognostic information. We found no significant difference in satisfaction with prognostic information at any time point. (Table 2 and Figure 3) Neither did we find any significant difference when stratifying by age.

Table 2.

Patients’ satisfaction with the amount of information they have about prognosis as reported at baseline, 4 months, and 12 months

	4 months
	I wish I had more information about my likelihood of cure.	I now have about the right amount of information.	Total
Baseline
I wish I had more information about my likelihood of cure.	10 (7)	16 (12)	26 (19)
I now have about the right amount of information.	9 (7)	99 (74)	108 (81)
Total	19 (14)	115 (86)	P=.16
	12 months
	I wish I had more information about my likelihood of cure.	I now have about the right amount of information.	Total
4 months
I wish I had more information about my likelihood of cure.	14 (10)	5 (4)	19 (14)
I now have about the right amount of information.	12 (9)	103 (77)	115 (86)
Total	26 (19)	108 (81)	P=.09

McNemar’s test of symmetry, relative to prior time point.

Figure 3. Patients’ Longitudinal Satisfaction with the amount of Prognostic Information.

No significant difference between time points when evaluated by McNemar’s test of symmetry, relative to prior time point.

Factors Associated with Satisfaction with Prognostic Communication

In bivariable analyses, patients who reported a high-quality physician communication style were more likely to be satisfied with the amount of information received about prognosis (OR 2.67, 95% CI 1.38 to 5.17, Table 3), as were those >75% chance of cure (OR 2.39, 95% CI 1.24 to 4.61). Results of multivariable analysis were similar, with prognostic satisfaction associated with high quality communication (OR 1.59, 95% CI 1.39 to 8.55) and more favorable prognosis (OR 4.03, 95% CI 1.58 to 10.27) after adjustment for patient age category, race/ethnicity, and time point of survey administration.

Table 3.

Factors Associated with Having the Right Amount of Prognostic Information - Bivariable Logistic Regression

	OR of having the right about of prognostic information (95% CI)
Patient Characteristics
Age (15-17)	Reference
Age (18-21)	.77 (.28-2.11)
Age (22-29)	.63 (.27-1.45)
White non-Hispanic race/ethnicity	.64 (.26-1.60)
Female gender	.57 (.31-1.06)
Relapse at 4 months	.60 (.08-4.49)
Relapse at 12 months	1.04 (.22-4.92)
>75% chance of cure at baseline as rated by physician	2.39 (1.24-4.61)
Communication Experiences
High quality communication	2.67 (1.38-5.17)
High quality information received at baseline	1.64 (.88-3.07)
Complete trust in oncologists	1.47 (.73-2.93)
Liking oncologists a great deal	2.00 (1.04-3.86)

Mixed linear models using random effects modeling for time. Bolding indicates p<.05.

In exploratory analyses, we performed bivariable logistic regression analyses to evaluate for potential psychosocial outcomes of satisfaction with prognostic information. (Supplemental Table S2) In these analyses, satisfaction with prognostic information was inversely associated with anxiety (OR 0.44, 95% CI 0.23 to 0.81), but not statistically associated with regret, depression, or peace of mind.

Discussion

In this longitudinal study, most AYA patients expressed a strong desire for prognostic information, beginning at time of diagnosis and increasing over the following year. While many AYA patients were satisfied with the information they received, a substantial minority wanted more information at each time point. Almost no patients ever desired less information, and few considered this information to be upsetting.

This strong preference for prognostic information is consistent with previous work showing that AYA patients generally wish to be well informed and desire a prominent role in decision making and care,²⁷ even at the end of life.²⁸ However, most prior work on cancer communication with AYA patients has been conducted at single time points during treatment, at the end of life, or in survivorship, without consideration of changes in information needs over time. Our longitudinal findings offer several points for consideration.

First, although prognosis is among the most difficult topics addressed during cancer treatment, these AYA patients generally wanted information at time of diagnosis; even more wanted information over time. In adult oncology, cancer patients experience initial “existential plight” (intense emotional distress, usually focused on life and death concerns and physical symptoms) during the first few months after diagnosis. This phase of illness often reflects a loss of meaning, and is usually followed by a period of “active coping” in which the patient tries to reestablish equilibrium and meaning.^{3, 29–31} These coping strategies may alter the patient’s response to prognostic information. AYAs similarly employ several coping strategies over time after an initial crisis.³² As young patients become accustomed to the diagnosis and prognosis of cancer, they might need prognostic information to address new questions that emerge, or they might grow in their readiness to hear and process difficult information. Oncologists should revisit conversations about prognosis over time, even (or especially) with patients who did not wish for this information initially.

Second, few AYA patients considered prognostic information upsetting, and distress was not associated with prognosis. Instead, distress and anxiety were associated with a desire for more information. Although clinicians might assume that emotional distress would indicate a wish to avoid prognostic information, patients who were upset by prognostic information wanted more information, not less. This can pose challenges to clinicians who hope to protect their patients from the harms of bad news. Although the process of considering one’s own mortality is upsetting, providing the opportunity to reflect on this information and discuss it openly is often therapeutic.

Third, a sizeable minority of patients wanted more prognostic information over the first year, and these patients gave lower ratings of physician communication style. This finding could indicate that satisfaction with communication is less about what is communicated and more about how it is communicated. Previous studies have found that the manner of disclosure greatly affects the way AYAs receive information.¹⁵ Our findings indicate that care and sensitivity positively affect perceptions of communication. For clinicians, it is important to recognize that the diagnosis of cancer and initiation of treatment is a major crisis for most patients. In a crisis, patients use defenses to allow themselves time to adapt to this new and difficult situation. As they settle into the reality of their new situation, defenses go down and patients likely have the capacity and desire for more information.^{30, 31}

We know from adult oncology that patients employ various coping strategies that can differ depending on age and clinical circumstances.^{3, 29, 33} We also know that seeking support and active coping strategies are most prevalent,^{3, 29, 33} and the use of denial or avoidance coping has been associated with increased severity of grief ³² and higher anxiety, depression, and cancer-related worries.^{29, 33} If these data apply to AYA patients, then most patients will employ active coping strategies over time and would benefit from ongoing prognostic disclosure, which our findings support. For patients who consistently prefer to avoid prognostic discussions, clinicians should not force this information, but they should consider whether these patients are at heightened risk of anxiety, depression, or worry. As such, clinicians could strongly encourage psychological evaluation and support for these patients.

Of note, we did not find large differences by age in either stratified or multivariable analyses, although we had expected that the youngest AYAs might be especially vulnerable to communication challenges. These subgroup analyses, however, had relatively small numbers, so lack of difference should be interpreted cautiously. Previous small, qualitative studies suggest that AYA patients with cancer desire developmentally appropriate provision of information,³⁴ underscoring the importance of continued study of age-specific needs.

This study had limitations worth consideration. First, this study was performed at a single, tertiary cancer center, and patients may have had higher information needs than the general population. Most patients also had favorable prognoses, and minorities were underrepresented. These factors could affect the generalizability of our findings. Our study was also observational, thus we cannot determine causality. Furthermore, we did not explore patients’ coping mechanisms, which could have provided important information. Nor did this study address physicians’ innate communication skills and self-efficacy, treatment intensity, or health locus of control, among many other psychosocial outcomes. Lastly, oncologists only estimated prognosis at baseline. Therefore, we cannot determine if prognostic estimates changed over time.

Many AYA patients with cancer in this study were satisfied with prognostic disclosure throughout the year after diagnosis, but a sizeable minority reported unmet information needs. Oncologists should understand the perceived value of prognostic information and the need to revisit these conversations over time. However, oncologists should also be sensitive to the unique and changing needs of AYA patients, and they should consider each patient to be the expert on his or her current needs, desires, hopes, and fears. Rather than making assumptions based on the patient’s age or presumed level of development, oncologists should seek each patient’s input to guide the disclosure process. By addressing these prognostic communication needs, clinicians have the potential to alleviate distress and anxiety while helping young people to cope with challenges that lie ahead.