Pediatric Advance Care Planning and Adolescent Preparedness and Quality of Life: An RCT

Abstract

BACKGROUND AND OBJECTIVE:

To evaluate the efficacy of Family-Centered Advance Care Planning for Teens With Cancer (FACE-TC) on adolescents’ quality of life.

METHODS:

A clinical trial randomized adolescent-family dyads at a 2:1 ratio to either FACE-TC or control. FACE-TC dyads received 3 weekly 60-minute sessions: Lyon Pediatric Advance Care Planning Survey; Next Steps: Respecting Choices; and Five Wishes. Generalized mixed-effect models evaluated efficacy at 3, 6, and 12 months after intervention measured by FACIT-SP-Ex-V4 (meaning/peace, faith) and PROMIS pediatric (anxiety; depressive symptoms; pain interference, fatigue). Fisher exact tests assessed decisional support and preparedness.

RESULTS:

Adolescents (n = 126) were mean age 17 years, 57% female, and 79% white. No significant differences were found between groups for faith or meaning/peace. At 12 months after intervention compared to control, FACE-TC increased anxiety (mean ratio 1.14; CI 1.04–1.25), depressive symptoms (mean ratio 1.12; CI 1.02–1.22), and pain interference (mean ratio 1.10; CI 1.00–1.20), but not at 3 or 6 months. FACE-TC increased fatigue at 3 months (mean ratio 1.13; CI 1.02–1.26), but not at 6 or 12 months. Compared to control, adolescents participating in FACE-TC agreed that “I feel prepared for the future” (76% vs 94%) and “I feel we are now on the same page” (76% vs 94%) at 3 months, but not at 12 months.

CONCLUSIONS:

There were no significant differences in quality of life between groups until 1 year, except for fatigue. FACE-TC had late effects, increasing adolescents’ anxiety, depressive symptoms, and pain interference. Reassessment at 1 year is clinically important.

INTRODUCTION

Pediatric advance care planning (pACP) is a family-centered decision-making process that involves discussion about goals of care and future medical care treatment preferences before a medical crisis.¹ In rigorous randomized clinical trials (RCTs), the Family-Centered Advance Care Planning for Teens With Cancer (FACE-TC) intervention has demonstrated (1) increased communication and agreement about end-of-life treatment preferences between adolescents with cancer^2,3 and HIV⁴ and their families for 1 year, (2) increased quality of life (QoL) for families,⁵ and (3) increased QoL for adolescents living with HIV with respect to disease-specific symptoms 12 months after intervention.⁶

The primary outcome (longitudinal congruence) for the current trial was already reported.³ This study focuses on secondary outcomes related to adolescents’ preparedness and QoL longitudinally. We hypothesized that adolescents participating in FACE-TC would report greater improvement in QoL than adolescents undergoing treatment as usual (TAU) over 12 months after intervention. We hypothesized that adolescents participating in FACE-TC would report greater decision-making support and preparedness than adolescents undergoing TAU over time.

METHODS

Study Design and Participants

We conducted a prospective, assessor-blinded, intent-to-treat, 2-arm RCT. Reported here are the results of the 3-, 6-, and 12-month postintervention assessments for adolescents only. We excluded the 18-month data, as our primary outcome showed no intervention effect after 12 months.³ Adolescent inclusion criteria were the following: (1) age 14 years or older and younger than age 21 years at enrollment; (2) not known to be developmentally delayed; (3) not severely depressed⁷; (4) speaks English as primary language; and (5) knows diagnosis. Secondary screening was performed for exclusion criteria—homicidality, suicidality, or active psychosis—to ensure competency to engage in decision-making. Appropriate referrals were given. Adolescents were recruited from 4 pediatric quaternary hospitals: Akron Children’s, Children’s National, St. Jude Children’s Research, and the University of Minnesota Masonic Children’s. Adolescents with all malignancies at any stage of illness and with all prognoses were eligible. Enrollment occurred between July 16, 2016, and April 30, 2019. Eligible adolescents underwent written informed consent/assent using site-specific institutional review board–approved documents prior to procedures and then completed the baseline assessment followed by randomization at the level of the adolescent-family dyad. An external safety-monitoring committee monitored the trial. We followed the Consolidated Standards of Reporting Trials checklist.

Procedure

Dyads were randomized to either the FACE-TC intervention or the TAU control group at a ratio of 2:1 (N = 83 FACE-TC dyads and N = 43 TAU control dyads). See Figure 1. Allocation was concealed from the research assistant (RA) assessor to prevent bias. Block randomization by study site controlled for site differences. Recruitment procedures and detailed methods have been previously reported.^2,3

FIGURE 1.

Flow of participants through the trial.

Abbreviations: ACP, advance care planning; F, family.

Intervention and control conditions

FACE-TC

FACE-TC consisted of 3 weekly 60-minute sessions. Prior to randomization, adolescents aged 18 to 21 years chose surrogate decision-makers to participate in the FACE-TC intervention with them. For adolescents aged 14 to 17 years, legal guardians participated. In Session 1, participants completed the Lyon ACP Survey separately, which primed them for issues discussed in Session 2.⁸ Session 2, Next Steps: Respecting Choices, enabled the adolescent and family to engage in a conversation with a certified facilitator about goals of care and treatment preferences.⁹ In Session 3, the adolescent and family completed the legal advance directive, Five Wishes.¹⁰ A copy was given to the family and the oncologist and placed in the medical record by the facilitator, consistent with pediatric oncology recommendations for structured pACP to facilitate communication and medical decision-making for adolescents with cancer.¹¹ See Supplemental Table 1for session goals and processes.

Treatment as usual comparison condition

We chose a TAU comparison condition to minimize patient burden. Both groups received an a pACP booklet.

Data Source and Measures

Measures were administered separately to the adolescent in a private room by the RA assessor, who was not the interventionist. Some follow-up assessments were completed by telephone. Assessors read the questions aloud and entered the responses immediately into the database through a laptop. Demographic data were collected on a standardized form. Reporting race and/or ethnicity was required by the funder.

Spiritual Well-Being Scale of the Functional Assessment of Chronic Illness Therapy version 4 (FACIT-Sp)¹² is a 23-item scale focusing on the existential aspects of spirituality as one aspect of QoL. We used 2 subscales: meaning/peace and faith. Higher scores indicated higher spiritual QoL. This measure has been used with adolescent outpatients with HIV and cancer.^12–14

Patient-Reported Outcomes Measurement Information System (PROMIS-SF version 1.1 8b^15–18) is a self-report developed for use with patients with pediatric cancer to assess health-related QoL. Emotional distress (anxiety), emotional distress (depressive symptoms), fatigue, and pain interference forms each have 8 items. Feasibility, acceptability, and validity are established.^15–18 We used PROMIS with participants aged 18 to 20 years, although norms are only available up to age 18 years. Interpretation of T-scores are 20 to 50 within normal limits, 50 to 55 (mild), 55 to 65 (moderate), and 65 to 85 (severe).¹⁹ Higher scores reffect greater endorsement of the construct compared to norms.²⁰ The minimally important difference (MID) is 3 points for adolescents.¹⁸

The Brief Multidimensional Measurement of Religiousness/Spirituality (BMMRS [adapted])^21,22 for use in health research assessed the moderating effects of 5 items shown to be important in our previous research on QoL.^23,24 Items were the following: the participant feels God’s presence; the participant prays privately; the participants goes to religious services; the participant identifies as religious; and the participant identifies as spiritual.

The Decisional Support and Preparedness Questionnaire(DSaP) was developed as a study-specific tool assessing longitudinal satisfaction with FACE protocol.²⁵ Responses were self-reported on a 5-point Likert scale with higher scores indicating more preparedness.

In a post hoc analysis, investigators added COVID-19 of March 1, 2020, as a dynamic time-dependent variable.

Statistical Analysis

Analysis was by original assigned groups using the intent-to-treat design (Figure 1). Data were entered into REDCap.²⁶ Descriptive statistics were computed. For the primary outcome QoL analysis, the adolescent sample size was N = 123 at baseline, N = 107 at 3 months, N = 108 at 6 months, and N = 104 at 12 months. Generalized linear mixed-effect models²⁷ with temporal autocorrelation within patient, random intercept per patient, gamma response, and logarithmic link function were used to model each outcome and to test the effect of the intervention on 6 outcomes: meaning/peace, faith, anxiety, symptoms of depression, pain interference, and fatigue at 3, 6, and 12 months after intervention, controlling for baseline levels. We considered 20 variables (time since diagnosis, intervention, 5 items from BMMRS, patient age, patient gender, parent employment status, parent annual income, family size, age class, under the 2016 federal poverty line [FPL], education, white, Black, Hispanic, indicator variable for time of assessment, and COVID-19) as covariates for each of the models, as prior research indicated that these variables, including self-identified race,^6,24,28 might inffuence outcomes. We only included the most significant variables from a practical, statistical, and standardized point of view.²⁹ Model variable selection was accomplished using L1-penalized estimation and cross-validation.³⁰ We kept the covariates age, gender, race, family income under the FPL, on/off treatment, and indicator for baseline visit or 3-, 6-, and 12-month follow-up in all models. There were no missing values for the outcomes or predictors that were used. Missing values were by attrition only. Residual plots were studied as goodness-of-fit tests. Using the fitted models, we computed appropriate contrasts and corresponding 95% CIs for each outcome to estimate the 16 pairwise comparisons between and among each of the different times of measurement and the 2 patient groups (Figure 2 and Supplemental Table 2).

FIGURE 2.

Estimated mean ratios for each of 6 quality-of-life outcomes using generalized mixed-effects models assessing the intervention effect.

Statistical significance was set to α = 0.05. The R statistical software was used for data analysis.³¹ The main R packages used for analysis, model development and validation, and table and plots generations are tidyverse,³² glmmTMB,³³ doParallel,³⁴ glmmLasso, flextable, and table-1.35 For rate ratio estimates and 95% CIs of effect of the intervention at different times of observation on the 6 outcomes, see Supplemental Table 2. For contrasts within patient groups over time for the 6 outcomes, see Supplemental Figure 1.

Differences in responses between groups for each question on the DSaP were assessed using the 2-sided Fisher exact test.

RESULTS

Participant Characteristics

Adolescents’ (N = 126) ages ranged from 14 to 20 years with a mean of 17; 57% were female, fewer than one-fourth were on therapy, and length of time since diagnosis ranged from 1 month to 19 years with a mean of 6 years. See Table 1.

TABLE 1.

Baseline Demographic and Clinical Characteristics of Adolescents with Cancer (n = 126)

Variable	FACE-TC (n = 83)	Treatment as Usual Control (n = 43)
Age, mean (SD) [range], y	16.9 (1.8) [14–20]	17.0 (2.0) [14–20]
Sex, n (%)
Female	45 (54.2)	27 (62.8)
Male	38 (45.8)	16 (37.2)
Race,a n (%)
Asian	3 (3.6)	0 (0)
Black or African American	12 (14.5)	5 (11.6)
White	63 (75.9)	37 (86.0)
More than 1 race	4 (4.8)	1 (2.3)
Declined to answer	1 (1.2)	0 (0)
Ethnicity, n (%)
Hispanic or Latino	5 (6.0)	0 (0)
Not Hispanic or Latino	76 (91.6)	40 (93.0)
Declined to answer	2 (2.4)	3 (7.0)
Parents’ educational level
No high school diploma or GED equivalency	2 (2.4)	0 (0)
High school or GED equivalency	16 (19.3)	7 (16.3)
Some college but no bachelor’s degree	31 (37.3)	17 (39.5)
College or advanced degree	34 (41.0)	19 (44.2)
Household income, FPL
Equal or below FPL	21 (25.3)	12 (27.9)
On active treatment?
Yes	20 (24.1)	7 (16.3)
No	63 (75.9)	36 (83.7)
Cancer diagnosis
Leukemia	26 (31.3)	16 (37.2)
Solid tumors	25 (30.1)	9 (20.9)
Brain tumor	17 (20.5)	8 (18.6)
Lymphoma	10 (12.0)	9 (20.9)
Otherb	5 (6.0)	1 (2.3)
Length of time since diagnosis, mean (SD) [range], mo	81 (68) [1–232]	70 (60) [2–198]

Abbreviations: FACE-TC, Family-Centered Advance Care Planning for Teens With Cancer; FPL, 2016 federal poverty level; GED, General Educational Development.

^aRace and ethnicity were self-reported by adolescents. Reporting race and/or ethnicity was required by the funder.

^bOther, totally resected World Health Organization grade II ependymoma, Ewing sarcoma, germ cell tumor, nonosseous retroperitoneal Ewing sarcoma, papillary thyroid carcinoma, plexiform neurofibroma, Wilm tumor, melanoma, and GATA2-related myelodysplastic syndrome, neuroblastoma, pilonidal abscess, synovial sarcoma of soft tissue, and metastatic melanoma.

The participation rate was 39% (260/660 participants or 130/330 adolescent-family dyads). Retention was 83% (104/126 fully eligible dyads) at 18 months after intervention.³ Four patients died of their disease: 2 intervention and 2 control. One intervention parent died. Attrition (including of the deceased) differed between groups with 20% (17/83) for intervention dyads and 12% (5/43) for control dyads. See Figure 1. No study-related adverse events were reported.

Pain Interference

There were no between-group differences in pain interference using the mean ratio (3 months = 1.04, 95% CI 0.95– 1.14, P = .35; 6 months = 1.07, 95% CI 0.98–1.17, P = .13; 12 months = 1.10; 95% CI 1.00–1.20, P = .05). The 12-month pain interference was at moderate levels for 13% (8/61) of adolescents participating in FACE-TC compared with 6% (2/36) of TAU and at severe levels only for adolescents participating in FACE-TC (1% [2/61] vs 0% [0/36]) (P = .08). See Supplemental Table 3. Twelve-month mean score differences exceeded the MID: 40 (SD = 7.99) vs 44 (SD = 10.18).¹⁸

Fatigue

FACE-TC increased fatigue at 3-month follow-up but not 6 or 12 months. Mean ratio at 3 months = 1.14, 95% CI 1.02–1.26, P = .02; at 6 months = 1.05, 95% CI 0.94–1.17, P = .38; at 12 months = 1.11, 95% CI 0.99–1.24, P = .06. Three-month fatigue was at moderate levels for 24% (16/61) of adolescents enrolled in FACE-TC compared with 16% (6/36) of those enrolled in TAU and at severe levels for 14% (9/61) of adolescents enrolled in FACE-TC compared with 5% (2/36) of those enrolled in TAU (P = .31). See Supplemental Table 3. Three-months mean score differences exceeded the MID: 42 (SD = 12.40) vs 45 (SD = 12.86).¹⁸

Emotional Distress: Anxiety

There were no between-group differences in anxiety at 3 or 6 months, but anxiety increased significantly for FACE-TC adolescents at 12 months. Mean ratio at 3 months = 1.05, 95% CI 0.96–1.14, P = .26; at 6 months = 1.01, 95% CI = 0.92–1.10, P = .79; at 12 months = 1.14, 95% CI 1.04–1.25, P = .004. See Figure 2 and Supplemental Table 2. Twelve-month anxiety scores were at moderate levels for 30% (18/61) of adolescents enrolled in FACE-TC compared with 14% (5/36) of those enrolled in TA, and anxiety scores were at severe levels only for adolescents enrolled in FACE-TC (12% [7/61] vs 0% [0/36], P = .01). See Supplemental Table 3. Twelve-month score differences exceeded the MID 44.42 (SD = 9.14) vs 51.63 (SD = 10.99).¹⁸

Emotional Distress: Depressive Symptoms

There were no between-group differences in depressive symptoms at 3 or 6 months, but there was a significant increase for adolescents enrolled in FACE-TC at 12 months. Mean ratio at 3 months = 1.04, 95% CI 0.95–1.14, P = .40; at 6 months = 1.07, 95% CI 0.98–1.18, P = .12; at 12 months = 1.12, 95% CI 1.02–1.22, P = .02. Twelve-month depressive symptoms were at moderate levels for 23% (14/61) of adolescents enrolled in FACE-TC compared to 11% (4/36) of those enrolled in TAU, and depressive symptoms were at severe levels only for adolescents enrolled in FACE-TC (12% [7/61] vs 0% [0/36], P = .046). See Supplemental Table 3. Twelve-month mean between-group scores exceeded the MID: 43 (SD = 8.25) vs 49 (SD = 10.99).¹⁸

Meaning/Purpose and Faith

No significant between-group differences were found for faith or meaning/peace (faith: mean ratio at 3 months = 0.92, 95% CI 0.73–1.16, P = .47; at 6 months 0.96, 95% CI = 0.72–1.22, P = .75; at 12 months 0.92; 95% CI = 0.72–1.17; P = .48; meaning/peace: mean ratio at 3 months = 0.98, CI 0.88–1.09, P = .76; at 6 months 0.97, CI = 0.87–1.08, P = .60; at 12 months 0.92; CI = 0.82–1.02; P = .10). See Figure 2 and Supplemental Table 2.

The demographic effects of age and family income under the 2016 FPL on QoL are provided in Supplemental Table 4.

Decisional Support and Preparedness

Compared to those undergoing TAU, adolescents participating in FACE-TC were significantly more likely to agree that they “feel prepared for the future” (76% vs 94%, P = .03) and that they “feel we are now on the same page” (76% vs 94%, P = .04) at 3 months, but not at 12 months. See Figure 3. Qualitative responses to “Is there anything else you want to tell us about how you feel about the FACE study?” are provided for adolescents participating in the intervention in Supplemental Table 3 for 3-, 6-, and 12-month follow-ups.

FIGURE 3.

Adolescent decisional support and preparedness at 3 (N = 107) and 6 (N = 104) months after intervention. Abbreviation: FACE-TC, Family-Centered Advance Care Planning for Teens With Cancer.

COVID-19

We conducted a post hoc analysis of adolescents who completed the 12-month assessment before COVID-19 (N = 59) vs after COVID (n = 5) to see if this might account for the QoL at the 12-month assessment. Results indicate that completion of the 12-month assessment after the outbreak of the COVID-19 pandemic resulted in a significant decrease in meaning/peace (mean ratio: −0.32; CI = −0.32 to −0.3; P = .02) and faith (mean ratio: −0.42; CI = −0.71 to −0.14; P = .004) compared 12-month assessments completed before the outbreak of COVID-19. COVID-19 had no effect on PROMIS outcomes, as shown in Supplemental Table 3.

DISCUSSION

Adolescent cancer survivors agreed to participate in a longitudinal, controlled, pACP trial with high retention rates. They engaged with their families in deep and vulnerable conversations about goals of care in the event of a future medical crisis.

Contrary to the hypothesis and pilot trial,³⁶ the intervention did not improve physical, emotional, or spiritual QoL for adolescent cancer survivors. There were no group differences in emotional (anxiety or symptoms of depression) or physical (pain interference) QoL at 3 or 6 months after intervention. Furthermore, 12-month symptoms of anxiety, depression, and pain interference increased significantly, both statistically and in clinically meaningful ways, for adolescents participating in the FACE-TC intervention compared with controls. Qualitative comments in Supplemental Table 3 provided no indicators for increased symptoms. FACE-TC also had no significant effect on peace/meaning at any point in time.

A post hoc analysis of COVID-19 effects showed significantly decreased peace/meaning and faith for intervention adolescents compared to controls. COVID-19 had no impact on symptoms, contrasting with surveyed teens with cancer who reported more anxiety after the pandemic³⁷ and a systematic review examining adolescents before and after COVID-19.³⁸ Perhaps different symptom questionnaires or our small sample account for the negative findings.

Consistent with pediatric participation in a medical decision-making framework^39,40 and the study hypotheses, FACE-TC intervention increased adolescents’ feeling of being prepared and supported for future medical decision-making and of being on the same page with their families. However, this effect was not sustained at 1 year, even though families accurately reported their adolescents’ treatment preferences at 1 year.³ Engagement in pACP contributed positively to short-term development of decision-making competence,⁴¹ consistent with research demonstrating that structured approaches to pACP facilitate communication between adolescents with cancer, their families, and treating clinicians.^3,6,42

These findings contradict a longitudinal trial of adolescents living with HIV using the FACE model.⁶ In that trial, HIV-specific symptoms decreased significantly at 1 year after intervention compared with controls.⁶ Religiousness and spirituality moderated adolescents’ health-related QoL, ie, higher religiousness was associated with higher symptom burden for all participants.⁶ This sample was 99% African American. In the adult trial, of whom 84% were African American, FACE-HIV had no effect on QoL.²⁴ Rather, participants were more likely to feel higher levels of emotional and physical well-being if they attended religious services regularly, prayed daily, felt God’s presence, and self-identified as religious or spiritual.²⁴ By contrast, participants who had the lowest levels of QoL were privately religious.²⁴ Song and et al found that the SPIRIT ACP intervention was more effective for adult African Americans than non-African Americans on dialysis in improving preparation for end-of-life decision-making and post-bereavement outcomes.²⁸

Alternatively, the difference in QoL between the control and treatment group may be due to the higher percentage of patients in the control group with leukemia/lymphoma than in the treatment group (58.1% vs 43.3%), which has better overall prognosis. Similarly, the intervention group was 11 months further from diagnosis, which may have impacted long-term outcomes.

The late effect of increased symptoms at 12 months could also be due to health-related existential uncertainty. Health-related existential uncertainty is defined as “an awareness of the undetermined but finite nature of one’s own being in-the-world, concerned primarily with identity, meaning, and choice.”⁴³ Intervention adolescents confronted the possibility of death directly, selecting a surrogate decision-maker, creating ACPs with their families, agreeing to put an advance directive in the medical record, and sharing results with their oncologist. Perhaps the developmental tasks of late adolescents and young adulthood⁴⁴ posed challenges related to identity, meaning, and choice beyond their capacity to cope. Qualitative responses to how they experienced the intervention provided no insights. Interventions have been developed and are being tested to help adolescent cancer survivors cope with these challenges.⁴⁵ A qualitative follow-up study with adolescents who participated in the intervention and whose symptoms worsened might help us to better understand what changed over time from their perspective. Additionally, clinical follow up at 1 year after pACP is warranted to assess QoL and spiritual concerns.

LIMITATIONS

The participation rate of 39% reffects the challenges of dyadic RCTs and may introduce selection bias and limit generalizability. Randomization corrects for potential bias. The participation challenges are similar to other pediatric oncology psychosocial intervention RCTs, where there are few examples of successful recruitment and retention strategies.⁴⁶ African American adolescent cancer survivors were underrepresented in this sample. Other limitations include the lack of non–English-speaking participants, although we later adapted the intervention for Spanish-speaking participants.⁴⁷ We did not assess for English proficiency. However, the RA assessor read all questions to all participants to control for literacy and understanding and entered the data into the REDCap database. Most adolescents were not on active treatment, although 5 patients died during the trial, and complications may arise later in life due to medical regimens and treatment side effects. There is lack of generalizability beyond academic pediatric hospitals. Our sample size after COVID-19 was small, so negative findings for symptoms may reffect a lack of power.

CONCLUSIONS

One year is a critical period to screen for QoL and psychosocial/emotional symptoms among adolescents who have participated in pACP with their families. It is unclear why there were late negative effects of FACE-TC on emotional and physical QoL. Trial findings support recommendations for skills-based, integrated psychosocial services in survivorship that serve both the mental and physical health of cancer survivors.^48,49 We exhort funders to allow researchers to conduct longitudinal natural history cohort studies of adolescent cancer survivors that extend at least 5 to 10 years after treatment. Such studies might help us to understand why QoL worsened over time for intervention adolescents and to design effective interventions to reduce suffering in this vulnerable population.

WHAT’S KNOWN ON THIS SUBJECT:

Without intervention, families and adolescents with cancer are often not on the same page about future medical care if the worst happened. FACE-TC significantly increased and maintained families’ knowledge of their adolescents’ treatment preferences and families’ quality of life.

WHAT THIS STUDY ADDS:

Longitudinal effects of advance care planning on quality of life and preparedness for adolescents with cancer are unknown. Adolescents participating in FACE-TC had significantly greater anxiety, depressive symptoms, and pain interference than controls at 12-month follow-up but not at 3- and 6-month follow-up.

ABBREVIATIONS

CI

Confidence Interval

FACE^® TC

Family-Centered Advance Care Planning for Teens With Cancer

pACP

pediatric advance care planning

QoL

quality of life

RCT

randomized clinical trial

TAU

treatment as usual