An Intervention in Congruence for End-of-Life Treatment Preference: A Randomized Trial

Abstract

BACKGROUND AND OBJECTIVES

There is a gap in family knowledge of their adolescents’ end-of-life (EOL) treatment preferences. We tested the efficacy of Family Centered Advance Care Planning for Teens with Cancer (FACE-TC) pediatric advance care planning (to increase congruence in EOL treatment preferences.

METHODS

Adolescents with cancer/family dyads were randomized into a clinical trial from July 2016 to April 2019 at a 2:1 ratio: intervention (n = 83); control (n = 43) to either 3 weekly sessions of FACE-TC (Lyon Advance Care Planning Survey; Next Steps: Respecting Choices Interview; Five Wishes, advance directive) or treatment as usual (TAU). Statement of Treatment Preferences measured congruence.

RESULTS

Adolescents’ (n = 126) mean age was 16.9 years; 57% were female and 79% were White. FACE-TC dyads had greater overall agreement than TAU: high 34% vs 2%, moderate 52% vs 45%, low 14% vs 52%, and P < .0001. Significantly greater odds of congruence were found for FACE-TC dyads than TAU for 3 of 4 disease-specific scenarios: for example, “a long hospitalization with low chance of survival,” 78% (57 of 73) vs 45% (19 of 42); odds ratio, 4.31 (95% confidence interval, 1.89–9.82). FACE-TC families were more likely to agree to stop some treatments. Intervention adolescents, 67% (48 of 73), wanted their families to do what is best at the time, whereas fewer TAU adolescents, 43% (18 of 42), gave families this leeway (P = .01).

CONCLUSIONS

High-quality pediatric advance care planning enabled families to know their adolescents’ EOL treatment preferences.

What’s Known on the Subject:

Gaps exist in a family’s knowledge of the end-of-life (EOL) treatment preferences of their adolescent with cancer. Evidence from a single-site pilot study supports the efficacy of the Family-Centered Advance Care Planning for Teens with Cancer (FACE-TC) intervention in increasing congruence in EOL treatment preferences.

What This Study Adds:

A 4-site, randomized clinical trial of FACE-TC closed the gap in families' knowledge. After FACE-TC pediatric advance care planning, and depending on the cancer-specific situation, families had 3 to 6 times the odds of accurately reporting their adolescents’ EOL treatment preferences than did controls.

There is a lack of discussion and documentation regarding end-of-life (EOL) care and treatment preferences for adolescents with serious illnesses.¹ Pediatric advance care planning (pACP) is an ongoing process of communication which aims to:

1. give adolescents a voice in medical treatment decisions if they are unable to speak for themselves;

2. prepare families so they know what their adolescent would want, if their child should experience serious illness complications;

3. communicate adolescents’ goals of care and EOL treatment preferences to clinicians; and

4. document these goals and preferences in the electronic health record.^2–4

Early pACP discussions can help seriously ill adolescents and families make future health decisions together about future medical treatments, if the adolescent could not communicate.^2–4 However, our team has documented that most families do not know what their adolescent with cancer or HIV would want, if the disease progressed.^5–8 In clinical practice, these adolescents rarely have documented advance care plans and the default is to provide intensive treatments that potentially increase suffering.^9,10 Despite cancer being the leading cause of disease-related death in adolescents,^11,12 conversations about goals of care and documentation of EOL care and treatment preferences for adolescents with cancer are not a routine and standard part of care.¹³

We developed and pilot tested a structured Family-Centered Advance Care Planning for Teens with Cancer (FACE-TC) in response to this gap in knowledge and documentation.^14–16 This developmentally and culturally appropriate intervention is informed by transactional stress and coping theory,¹⁷ Leventhal’s theory of self-regulation,¹⁸ and the representational approach to patient education.¹⁹ Surrogate decision-makers are hereafter referred to as “family.” A single-site pilot trial of FACE-TC demonstrated feasibility, acceptability, safety, and initial efficacy to increase congruence between adolescents and their families regarding EOL treatment preferences, documentation of pACP goals of care, and patients’ and families’ quality of life.^14–16 Our goal was to determine if FACE-TC could close the gap in knowledge of EOL treatment preferences between adolescents with cancer and their families in a large, multisite trial. We hypothesized that FACE-TC dyads would achieve better congruence in EOL treatment preferences than controls.

Methods

Study Design and Participants

Full details of the trial have been published.²⁰ The study design was a dyadic, (adolescent/family) 2-armed, parallel groups, single-blinded (assessor), intent-to-treat, controlled, randomized clinical trial (RCT). Eligibility criteria for adolescents were diagnosis of cancer, aged 14 to <21 years at enrollment, able to cognitively engage as determined by their oncologist, English-speaking, knows own diagnosis, and not in foster care. Family eligibility criteria were legal guardian of adolescent if adolescent was younger than 18 years, or chosen surrogate if adolescent was ≥18 years at enrollment, English-speaking, and knows patient’s diagnosis. After enrollment, participants underwent secondary screening for exclusion criteria, which included severe depression,²¹ homicidality,²² suicidality,²¹ and/or psychosis.²² Excluded participants were linked with psychological services. Dyads were recruited between July 16, 2016, and April 30, 2019, from 4 quaternary pediatric hospitals: Akron Children’s Hospital, Children’s National Hospital, St. Jude Children’s Research Hospital, and University of Minnesota Masonic Children’s Hospital.

The institutional review board at each site approved the protocol. Participants provided written informed consent or assent and received compensation of $25 per study visit. An external safety monitoring committee monitored the trial. This RCT follows the Consolidated Standards of Reporting Trials guidelines.

Procedures

Researchers completed 3 days of training on the protocol and annual booster sessions to ensure fidelity to the study protocol. Facilitators were nurses who received certification in the Respecting Choices training program.^23,24 The last author (M.L.) reviewed all videotapes of the Respecting Choices interview, using the Respecting Choices competency criteria checklist,^23,24 and provided supervision to facilitators during monthly conference calls.

After consulting with a patient’s primary oncologist, research assistants approached potentially eligible participants face-to-face during hospital outpatient visits and stays. The first visit included enrollment, secondary screening, and, if eligible, completion of baseline questionnaires. Randomization of dyads was at a 2:1 ratio, FACE-TC to treatment as usual (TAU), for ethical considerations, because the pilot^14,15 and earlier trials using the FACE model showed benefit.^25–28 We used a computer-generated randomization with 2 factors (ie, site and intervention arms), triggered by completion of baseline assessment. Facilitators notified the participants of random assignment to protect the blindness. Blinded assessors, who were not the interventionists/facilitators, read the questionnaires aloud to participants to control for issues of literacy or uncorrected vision, to monitor emotional reactions, to engage participants, and to ensure data completion. Questionnaires were administered to participants separately and in private.

FACE-TC pACP Intervention

Three sessions of ∼45 to 60 minutes were conducted weekly.

Session 1

The Lyon ACP Survey - Adolescent & Surrogate version is a 31-question survey which assessed the adolescents’ goals, values, experiences with death and dying, and EOL treatment preferences.⁶ The family version asked, for example, “What do you think (child’s name) thinks is the best timing for end-of-life discussions?” The adolescent and family member were administered the survey separately.

Session 2

The Next Steps: Respecting Choices^23,24 structured interview was adapted for adolescents with cancer.¹⁵ The adolescent and family engaged in a video-taped facilitated conversation, not completion of a form. The adolescent’s understanding of their cancer, complications that could occur, fears, hopes, and experiences with hospitalization and death were explored. Then, 4 cancer-specific situations were used to encourage discussion about the adolescent’s goals and values in “bad outcome” situations. They were asked to explain why they made their choices. The family was asked if they could honor the adolescent’s treatment preferences. The adolescent was then asked if they wanted their family to strictly follow their wishes or to do what they think is best in the moment. The facilitator noted adolescents’ questions on a post card, encouraging the adolescent to discuss their questions with their clinician at the next visit.

Session 3

The adolescent and family together completed the advance directive, Five Wishes.²⁹ The facilitator read each section of the form aloud. The facilitator gave a copy to the family, the primary provider by e-mail, and placed a copy in the electronic health record. For adolescents aged <18 years, their legal guardian’s signature was required on the Five Wishes advance directive. See Supplemental Table 5 for sessions’ foundations, goals, and processes.

Treatment as Usual Control

All participants were given a pACP booklet at the baseline assessment, but control dyads did not receive the 3-session–facilitated conversation. Assessments were administered at the same 2 time points as the intervention dyads (baseline and 3 weeks postbaseline) (Fig 1).

FIGURE 1.

Consolidated Standards of Reporting Trials Guidelines Checklist diagram of the FACE-TC trial using an intent-to-treat design.

Outcome Measure

The Statement of Treatment Preferences^23,24 documented the adolescents’ treatment preferences and obtained the family’s knowledge of the patient’s goals. Four cancer-specific situations and the benefits and burdens of treatment options were discussed:

1. long hospitalization with low chance of survival;

2. treatment would extend my life by no more than 3 months and side effects are serious;

3. physical impairment (cannot walk or talk and need 24-hour nursing care); and

4. cognitive impairment (do not know who you are, where you are, or who you are with and need 24-hour nursing care).

Choices were “to continue all treatments …,” “to stop all efforts to keep me alive…,” or “do not know.”

A follow-up question clarified the degree of leeway in decision-making authority the adolescent wished to grant the family member, that is, “I would want the person I have chosen to [either] “strictly follow my wishes” or “do what they think is best at the time, considering my wishes.” It was administered either after session 2 for intervention or at 3-weeks postbaseline for TAU.

A demographic data form was used to gather participant-reported age, sex, and race or ethnicity; also, family-reported education, employment status, household income, and household size. Chart abstraction data included diagnosis, date of diagnosis, history of relapse, history of bone marrow transplantation, and on or off active treatment.

Statistical Analysis

We first assessed the difference in the level of overall congruence in EOL treatment preferences in all 4 cancer-related situations between adolescents with cancer and their families. Dyadic responses were matched and a dichotomous variable (Yes: agreement; No: no agreement) was generated for each situation. Then, a 3-level categorical overall congruence measure was created: low (agree 0–1 situation), moderate (agree 2–3 situations), and high (agree all 4 situations). Fisher’s exact test was used to determine the association between the intervention and overall congruence. The strength of the association (effect size) was measured by Cramér’s V.³⁰

Pearson’s χ² or Fisher’s exact test was used to determine differences in dyadic congruence between the intervention and control groups in each of the cancer-related situations. Bonferroni correction was applied to adjust for the multiple testing. Odds ratios (OR) measured the effect size. See the Supplemental Information for Power Analysis.

Data were entered into Research Electronic Data software (version 8.10.18 ©2020). Statistical significance was set to α = 0.05. Analyses were conducted using SAS statistical software, version 9.4 (SAS Institute Inc.).

Results

Participant Characteristics

Of the 130 enrolled dyads, 4 did not meet inclusion criteria; thus, 126 dyads were fully enrolled for randomization (Fig 1). Of those who declined, 23% (46 of 203 dyads) had at least 1 member of the dyad who reported not wanting to talk about pACP, and 37% (76 of 203) had time issues. Compared with female adolescents, male adolescents were significantly more likely to decline participation (difference of 14%, 95% confidence interval [CI], 4–25, P = .02). Age, race, ethnicity, diagnosis, and active treatment status did not significantly differ between those who enrolled and those who declined.

Adolescents had a mean (SD) age of 16.9 (1.8) years, 57% (69 of 126) were female, and 79% (100 of 126) were White (Table 1). Most common diagnoses were leukemia (42 of 126, 33%), solid tumor (34 of 126, 27%), brain tumor (25 of 126, 20%), and lymphoma (19 of 126, 15%). Among families, more than half were at <200% of the 2016 federal poverty level, and 58% had less than a college education. Family members ranged in age from 19 to 67 years; 83% were female; 82% were White; 75% mothers; 15% fathers; and 8% (10 of 126) were nonbiological, patient-chosen surrogate decision-makers. Three adolescent patients had an advance directive in the medical record at baseline. Of the FACE-TC dyads who began Session 1, 90% (72 of 80) completed all 3 sessions (Fig 1). Ninety-one percent (115 of 126) completed all study visits.

TABLE 1.

Baseline Demographic Characteristics of Adolescents with Cancer and Their Families (N = 252 Participants)

Baseline Characteristics	Adolescentsa		Familiesa
	FACE-TC (n = 83)	TAU (n = 43)	FACE-TC (n = 83)	TAU (n = 43)
Age, y
Mean (SD)	16.9 (1.8)	17.0 (2.0)	45.6 (8.2)	46.5 (8.4)
Range	14–20	14–20	19–67	20–63
Sex
Female	45 (54)	27 (62.8)	67 (80.7)	37 (86.0)
Male	38 (46)	16 (37.2)	16 (19.3)	6 (14.0)
Race
American Indian or Alaska Native	0 (0)	0 (0)	0 (0)	1 (2.3)
Asian American	3 (3.6)	0 (0)	3 (3.6)	0 (0)
Black or African American	12 (14)	5 (12)	10 (12)	4 (9)
White	63 (76)	37 (86)	68 (82)	35 (81)
More than 1 race	4 (5)	1 (2)	2 (2)	3 (7)
Declined	1 (1)	0 (0)	0 (0)	0 (0)
Ethnicity
Hispanic or Latino	5 (6)	0 (0)	4 (5)	0 (0)
Not Hispanic or Latino	76 (92)	40 (93)	79 (95)	42 (98)
Declined	2 (2)	3 (7)	0 (0)	1 (2)
Education
No high school diploma or GED equivalency	45 (54)	26 (60)	2 (2)	0 (0)
High school or GED equivalency	28 (34)	8 (19)	16 (19)	7 (16)
Some college but no bachelor’s degree	9 (11)	9 (21)	31 (37)	17 (40)
Bachelor's degree, master's degree, doctorate, or professional degree	0 (0)	0 (0)	34 (41)	19 (44)
Declined	1 (1)	0 (0)	0 (0)	0 (0)
Income
Equal to or below the FPL	N/A	N/A	21 (25)	12 (28)
101%–200% of FPL	N/A	N/A	23 (28)	14 (33)
201%–300% of FPL	N/A	N/A	14 (17)	5 (12)
>300% of FPL	N/A	N/A	23 (28)	10 (23)
Declined	N/A	N/A	2 (2)	2 (5)

FPL, federal poverty line; GED, general equivalency diploma; N/A, not applicable.

^aData shown indicate n (%), unless otherwise indicated.

Congruence in EOL Treatment Preferences

The relationship of the family member to the patient was not significantly associated with congruence. Biological fathers (15 of 16, 94%) more accurately reported their adolescent’s treatment preferences, followed by biological mothers (61 of 87,70%), and then nonbiological family (7 of 12, 58%), as shown in Supplemental Tables 6 and 7. Congruence frequencies for each situation are shown in Supplemental Table 8.

Overall treatment preference congruence among adolescent/family dyads in 4 cancer-related situations was significantly greater in the FACE-TC group than that in the TAU group (P < .0001) (Table 2). Thirty-four percent (25 of 73) of intervention dyads had high agreement versus 2% (1 of 42) of controls; 52% (38 of 73) of intervention dyads had moderate agreement versus 45% (19 of 42) controls; and 14% (10 of 73) of intervention dyads had low agreement versus 52% (22 of 42) of controls. Cramér’s V was 0.48, indicating a moderate-effect size.

TABLE 2.

Statement of Treatment Preferences Agreement Group by Intervention (N = 115 Dyads)

Levels of Agreementa	FACE-TC pACP, Plus pACP Information Frequency (%)	TAU Control, Plus pACP Information Frequency (%)	Total
Agree 0–1 situation/ low agreementb	10 (14)	22 (52)	32
Agree 2–3 situations/ moderate agreement	38 (52)	19 (45)	57
Agree 4 situations/ high agreement	25 (34)	1 (2)	26
Total	73	42	115

Two-sided Fisher’s exact test, P < .0001. Cramér's V = 0.48. The strength of the association (effect size) was measured by Cramér's V, which has values ranging from 0 to 1: low association (V = 0.1–0.3), moderate association (V = 0.3–0.5), and high association (V > 0.5).²² Cohen's W is equivalent to Cramér's V in our study with 3 × 2 contingency table. The sample size was 115 dyads; 1 dyad had missing data.

^aRationale for recoding the data at 3 levels of agreement to generate a new variable is illustrated by the small number of cells for each situation.

^bDyadic “do not know” responses were treated as no agreement/no agreement, that is, low agreement.

Table 3 shows congruence by each cancer-specific situation. Results demonstrated statistically significant greater congruence for the intervention than controls for the following:

1. “a long hospitalization with low chance of survival,” 78% (57 of 73) vs 45% (19 of 42), OR, 4.31 (95% CI, 1.89–9.82);

2. “having only 3 more months to live and side effects of treatment are serious,” 77% (56 of 73) vs 36% (15 of 42), OR, 5.93 (95% CI, 2.58–13.63);

3. “cannot walk or talk and need 24-hour nursing care,” 63% (46 of 73) vs 33% (14 of 42), OR, 3.41 (95% CI 1.53–7.57); and

4. “do not know who you are or who you are with and need 24-hour nursing care,” 66% (48 of 73) vs 43% (18 of 42), OR, 2.56 (95% CI 1.17, 5.58). However, after making the Bonferroni correction, the group difference in situation 4 (cognitive disability) became statistically insignificant.

TABLE 3.

Overall Congruence of Statement of Treatment Preferences Postsession 2 or 3 Weeks Postbaseline Assessment by Study Arm (N = 115 Dyads)

Situations	Congruencea	FACE-TC pACP (n = 73)	TAU with pACP Information (n = 42)	P	OR (95% CI)
1. Long hospitalization with low chance of survival	No	16 (22)	23 (55)	.0003	4.31 (1.89–9.82)
	Yes, agree	57 (78)	19 (45)
2. Having only 3 more months to live and treatments	No	17 (23)	27 (64)	<.0001	5.93 (2.58–13.63)
have serious side effects	Yes, agree	56 (77)	15 (36)
3. Physical disability: cannot walk or talk and need 24-h nursing care	No	27 (37)	28 (67)	.002	3.41 (1.53–7.57)
	Yes, agree	46 (63.0)	14 (33.3)
4. Cognitive disability: do not know who you are, where you are, or who you are with and need 24-h nursing care	No	25 (34)	24 (57)	.0168b	2.56 (1.17–5.58)
	Yes, agree	48 (66)	18 (43)

The sample size was 115 dyads; 1 dyad had missing data.

^aDo not know responses were counted as no agreement/congruence.

^bAfter making the Bonferroni correction, the difference among groups in situation 4 was insignificant (corrected P = .0125). This is because our sample size (n = 115) achieved only 78% power to detect an OR of 2.56 in this situation.

Compared with TAU dyads, FACE-TC dyads were significantly more likely to agree to stop treatments in 2 situations (Table 4): “having only 3 more months to live and side effects of treatment are serious,” 56% (41 of 73) vs 21% (9 of 42), OR, 4.70 (95% CI, 1.97–11.21) and “do not know who you are or who you are with and need 24-hour nursing care,” 52% (38 of 73) vs 26% (11 of 42), OR, 3.06 (95% CI, 1.34–7.00). After making the Bonferroni correction, the group difference in “having only 3 months to live” became insignificant.

TABLE 4.

Congruence on Stopping Treatments, as Determined at Postsession 2 for Intervention or 3 Weeks Postbaseline for TAU by Study Arm (N = 115 Dyads)

Situation	Congruencea	FACE-TC pACP With Information (n = 73)	TAU With pACP Information (n = 42)	P b	OR (95% CI)
1. Long hospitalization with low chance of survival	No	55 (75)	39 (93)	.0235	4.25 (1.17–15.45)
	Yes, agree to stop	18 (25)	3 (7)
2. Having only 3 more months to live and treatments have serious side effects	No	32 (44)	33 (79)	.0003	4.70 (1.97–11.21)
	Yes, agree to stop	41 (56)	9 (21)
3. Physical disability: cannot walk or talk and need 24-h nursing care	No	60 (82)	37 (88)	.4015	1.60 (0.53–4,86)
	Yes, agree to stop	13 (18)	5 (12)
4. Cognitive disability: do not know who you are, where you are, or who you are with and need 24-h nursing care	No	35 (48)	31 (74)	.007	3.06 (1.34–7.00)
	Yes, agree to stop	38 (52)	11 (26)

The sample size was 115 dyads; 1 dyad had missing data.

^aDo not know responses were counted as no agreement/congruence.

^bAfter making the Bonferroni correction, the difference among groups in situation 1 was insignificant (corrected P = .0125).

Leeway

Adolescents were asked how strictly they wanted their families to follow their wishes. Adolescents in the FACE-TC group were more likely than those in the TAU group, 67% (48 of 73) vs 43% (18 of 42) (P = .01), to endorse “Do what he or she thinks is best at the time, considering my wishes.”

Discussion

To our knowledge, this is the first fully powered RCT to focus on adolescents with cancer and their engagement with their families in pACP conversations.^1,31–34 Findings confirmed our hypothesis that FACE-TC adolescent/family dyads achieved higher overall congruence than TAU control dyads in all 4 cancer-specific situations. Almost half of controls and slightly more than half of intervention dyads had moderate congruence, but only 1 control dyad and 25 FACE-TC dyads had high congruence in all 4 situations. In the control group, half (52%) of adolescents/families had low congruence, compared with only 14% of adolescents/families in the intervention group. Thus, FACE-TC adolescent/family dyads moved toward closing earlier identified congruence gaps.⁷

Compared with TAU dyads, FACE-TC dyads were also more likely to stop treatments in some scenarios, confirming our pilot study findings¹⁵ and HIV adolescent trials.^26,27 Thus, high-quality pACP communication fostered understanding in ways that allowed the adolescent and family choice in care trajectories. Allowing choice and giving permission to focus on quality of life may potentially mitigate the regret and distress of bereaved families who wished they had pursued more treatment of their child.³⁵ Control dyads agreed to stop treatments at higher frequencies in the current trial, compared with our single-site pilot, where no dyad in the control group agreed to stop treatments in situations involving a long hospital stay or only 3 more months to live. Perhaps the past 10 years represents a cultural shift in willingness to consider stopping treatments, with increasing national attention on the burdens of intensive treatments at the EOL.^36,37 With increased access to palliative care services³⁸ and research,^7,38,39 families and patients may better understand that stopping intensive medical interventions when the patient is dying is not giving up, but rather choosing how best to spend the final days of one’s life.

Adolescents in the FACE-TC group gave their families permission to do what they thought was best knowing their wishes, rather than strictly follow their wishes, replicating previous findings using the FACE model.^15,27,28,40 Granting leeway has important implications for bereavement care because regret and guilt have an impact on the intensity and complexity of grief and bereavement.⁴¹ In a multisite survey study, 73% of bereaved parents endorsed regret, often around treatment decisions.³⁵ Mack and colleagues found regret was less likely when bereaved parents felt satisfied with their role in treatment decision-making.⁴² Families knowing that their adolescent gave them leeway to do what is best in the moment may be quite important for bereavement care, especially in situations where the adolescent is no longer able to voice preferences. This is not well documented in the current literature.⁴³ Engagement of families in pACP did increase positive appraisals of their caregiving.⁴⁴ Future research should examine if pACP helps bereaved families cope with their grief, as has been found in an adult study of advance care planning (ACP).⁴⁵

Of those who declined, 23% (46 of 203) reported not wanting to talk about pACP. In some cases, the adolescent was ready, but the family was not.⁴⁶ In the FACE model assessment of readiness, to engage in pACP rests with the patient/family dyad, not the clinician,⁴⁷ thereby preventing gatekeeper bias (eg, mistakenly thinking the patient or family is not ready to discuss these topics) and honoring equipoise (ie, all adolescents with a serious illness have equal access to participation in this process).⁴⁸ To prevent family-level gatekeeper bias, other evidence-based models of ACP^49,50 could be offered for young adults aged 18 years or older whose family is not ready.

This study has limitations. Some analyses had low cell sizes. However, these were a direct reflection of the problem; for example, only 1 dyad in the control group had high congruence, likely not correctable with a larger sample. Thus, CIs were wide. There are no psychometric data on the Statement of Treatment Preference, although it is widely used, enabling replication of research findings.⁵¹ The participation rate of 39% affected generalizability, but is comparable to dyadic longitudinal trials of palliative care.^52,53 Talking about death and dying is taboo.⁵⁴ Some families believe it is their role alone to make EOL health care decisions or believe pACP is against their religion.⁵⁵ Study participants may represent early adopters or pioneers for pACP, potentially increasing its acceptance in the future. Generalizability may not extend beyond pediatric hospitals. Including those who had advance directives is consistent with recommendations for ongoing pACP conversations.^13,55,56 Social desirability bias could have occurred with face-to-face assessments. Although using hypothetical situations may not mirror real-life decision-making when the time comes, this likely is the first time such direct conversation has been held with adolescents. Like disaster preparedness,⁵⁷ this approach is valid, reducing people’s risk and increasing their ability to cope.^{14,15,25,44,45,58}

Conclusions

FACE-TC effectively increased communication between adolescents with cancer and their families about the patients’ EOL treatment preferences, meeting the first challenge of ACP, that is, knowledge of patient preferences.⁵⁹ This low-tech intervention commits to more deeply respecting adolescents and integrating them into health care decision-making.⁵⁶ Busy clinicians may benefit from this standardized and structured approach, which increased the odds that families knew what their child wanted and ensured the first conversation about goals of care and EOL treatment preferences did not occur during a medical crisis or in the ICU.

Glossary

ACP

advance care planning

CI

confidence interval

EOL

end of life

FACE-TC

Family-Centered Advance Care Planning for Teens With Cancer

OR

odds ratio

pACP

pediatric advance care planning

RCT

randomized clinical trial

TAU

treatment as usual