Patient Navigator Intervention to Improve Palliative Care Outcomes for Hispanic Patients With Serious Noncancer Illness

This randomized clinical trial aims to determine the effectiveness of a lay patient navigator intervention in improving palliative care outcomes among Hispanic patients.

Key Points

Question

Can a bicultural, bilingual lay patient navigator intervention improve palliative care outcomes for Hispanic patients experiencing serious noncancer illness?

Findings

In this randomized clinical trial of 209 Hispanic adults with serious noncancer illness, patients who received patient navigator visits did not have improved quality of life at 3 months, the primary study outcome, compared with patients who simply received educational materials. Patients in the intervention group did have improved advance care planning and increased hospice utilization, but pain and symptom outcomes were not different between the groups.

Meaning

A culturally tailored patient navigator intervention can help improve some palliative care–related outcomes for Hispanic patients with serious medical illness.

Abstract

Importance

Disparities persist across the trajectory of serious illness, including at the end of life. Patient navigation has been shown to reduce disparities and improve outcomes for underserved populations.

Objective

To determine the effectiveness of a lay patient navigator intervention, Apoyo con Cariño, in improving palliative care outcomes among Hispanic patients.

Design, Setting, and Participants

This was a multicenter randomized clinical trial that took place across academic, nonprofit, safety-net, and community health care systems in urban, rural, and mountain/frontier regions of Colorado from January 2017 to January 2021. Self-identifying Hispanic adults with serious noncancer medical illness and limited prognosis were recruited. Data were collected and analyzed from July 2022 to July 2023.

Interventions

Participants randomized to the intervention group received 5 home visits from a bilingual, bicultural lay patient navigator; participants randomized to control received care as usual. Both groups received culturally tailored educational materials. Investigators/outcome accessors remained blinded to participant assignment.

Main Outcomes and Measures

Change in score from baseline to 3 months on the Functional Assessment of Chronic Illness Therapy (FACIT) General quality of life (QOL) scale (primary outcome), Advance Care Planning (ACP) Engagement Survey, Brief Pain Inventory, Edmonton Symptom Assessment Scale, and FACIT Spiritual Well-Being subscale; at 6 months, advance directive (AD) documentation; and at 46 months or death, hospice utilization and length of stay, as well as aggressiveness of care at end of life.

Results

Of 209 patients enrolled (mean [SD] age, 63.6 [14.3] years; 108 [51.7%] male), 105 patients were randomized to control and 104 patients to the intervention. There were no statistically significant differences in the change in mean (SD) QOL score between the intervention and control groups (5.0 [16.5] vs 4.3 [15.5]; P = .75). Participants in the intervention group, compared with the control group, had statistically significant greater increases in mean (SD) ACP engagement (0.8 [1.3] vs 0.1 [1.4]; P < .001) and were more likely to have a documented AD (62 of 104 [59.6%] vs 28 of 105 [26.9%]; P < .001). There were no statistically significant differences in mean (SD) change in pain intensity score (0-10) between patients in the intervention group compared with control (−0.4 [2.6] vs −0.5 [2.8]; P = .79), nor pain interference (−0.2 [3.7] vs −0.4 [3.7]; P = .71). Patients receiving the intervention were more likely to be referred to hospice compared with patients receiving control (19 of 43 patients [44.2%] vs 7 of 33 patients [21.2%]; P = .04) and less likely to receive aggressive care at end of life (27 of 42 patients [64.3%] vs 28 of 33 patients [84.8%]; P = .046).

Conclusion and Relevance

In this randomized clinical trial, a culturally tailored patient navigator intervention did not improve QOL for patients. However, the intervention did increase ACP engagement, AD documentation, and hospice utilization in Hispanic persons with serious medical illness.

Trial Registration

ClinicalTrials.gov Identifier: NCT03181750

Introduction

Racial and ethnic minority patients experience disproportionate distress at the end of life. Hispanic adults are more likely to die in the hospital, which has been shown to be associated with increased pain, decreased quality of life (QOL), and increased posttraumatic stress disorder and complicated grief for family members.¹ Ethnicity is also increasingly recognized as a predictor for poor pain assessment and management.^2,3,4,5 Barriers to adequate pain management have been identified at the institutional level, due to bias, and as a result of diverse cultural beliefs.^6,7,8 While nationwide averages of completed advance directives (ADs) are low for all groups, Hispanic individuals are less likely to have a living will, medical durable power of attorney (MDPOA), or a do-not-resuscitate order^{9,10,11,12,13,14}; are more likely to choose very aggressive care in the face of serious or incurable illness; and are less likely to acknowledge their terminally ill status.^9,10 Overall, Hispanic persons are less likely to access hospice services.^15,16,17

A palliative approach is needed in advanced noncancer illnesses, given poor prognoses and high symptom burden.^18,19,20 Yet, specialty (tertiary)-level palliative care (PC) cannot grow fast enough to meet the demand.^21,22,23 Care models that promote primary PC, where a palliative approach is incorporated by the patient’s primary or specialty care physician (eg, cardiologist, pulmonologist), are required to maintain capacity for ever-growing needs, especially in poor urban and rural settings where tertiary PC is often nonexistent.^24,25,26,27 Lay patient navigators work with patients and families to reduce barriers to care, including educational, cultural, socioeconomic, and logistic barriers. Lay patient navigation models with an established evidence base in addressing health disparities^{28,29,30,31,32,33,34} have been adapted to help patients advocate for primary PC. Our research team demonstrated feasibility and acceptability of the lay patient navigator intervention for seriously ill Hispanic adults.³⁵ Previously, we conducted a randomized clinical trial of the lay patient navigator intervention for Hispanic adults with advanced cancer, finding improved rates of advance care planning (ACP), while pain remained low across both groups and hospice utilization was high in all participants.³⁶ This study’s goal was to determine the effectiveness of the navigator intervention to improve PC outcomes for Hispanic persons with serious noncancer medical illness. We hypothesized that the intervention group, randomized to receive the navigator intervention, Apoyo con Cariño, would have improved QOL (primary outcome), increased ACP, improved pain and symptom burden, increased hospice utilization, and decreased aggressive care at end of life compared with those who received usual care.

Methods

Sample

Participants were recruited from diverse outpatient settings consisting of nonprofit, community, safety-net, and academic health care systems across Colorado. Patients were screened for study eligibility (≥18 years old, self-identified as Hispanic, spoke either English or Spanish as a primary language, and not enrolled in hospice). Patients were eligible if they met prognostic criteria for life-limiting, serious noncancer illness as determined by 1 of 2 validated methods: (1) referred by a health care professional who answered no to the “surprise” question (“Would you be surprised if this patient died in the next year?”) or (2) meeting 1 of the noncancer CARING criteria (admitted ≥2 times in the past year for chronic illness diagnosis, residence in nursing home, intensive care unit admission with multiorgan failure, or meeting ≥2 of the National Hospice and Palliative Care hospice guidelines for noncancer illness) at the time of a recent hospital admission.^{37,38,39,40,41} Recruitment processes and procedures have been previously reported.⁴²

This study was approved by the Colorado Multiple Institutional Review Board, and all participants provided written informed consent. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.

Randomization

Patients were assigned to control or intervention group, stratified by site, using a 1:1 computer-generated blocked randomization with varying block sizes (2-6) following consent and baseline data collection. Patient navigators and the project manager (D.M.K.) were unblinded on group assignment; the primary investigator (S.M.F.) and all coinvestigators (K.L., W.G., R.M.F.) remained blinded throughout the study.

All participants (control and intervention) received an educational packet in English or Spanish, written at a fifth-grade reading level, that included booklets on ACP, pain and symptom management, hospice care, and a study-specific AD developed for a low health literacy population.⁴³ These materials were developed using surface and deep cultural tailoring. An interdisciplinary bicultural community advisory panel helped to select, modify, and adapt (culturally tailored) study materials.^35,44 Participants in the control group received usual care.

Intervention

In addition to written materials, patients randomized to the intervention also received home visits from a bilingual, bicultural patient navigator who lived in and served their local Colorado communities (Denver, Fort Collins/Greeley, mountain/frontier). All had health-related backgrounds (community health, home health, community-based research, and/or hospital-based nursing assistant). Details of the intensive 3-month training process, ongoing continuing education, and study protocol (Supplement 1) have been previously published.^35,45,46

The intervention, Apoyo con Cariño, was designed to include 5 navigator-initiated home visits for each intervention participant. Home visits followed a visit guide/script that served as a framework to ensure that all navigators were congruent in information delivery. Navigators reviewed educational materials and addressed barriers to 3 study domains: (1) ACP, (2) pain and symptom management, and (3) hospice utilization. The navigator used advocacy, activation, empowerment, education, and motivational interviewing to address barriers to the domains of interest^28,47 and was available by telephone or to make additional visits at the patient or family caregiver’s request.

Effect of the COVID-19 Pandemic

In March 2020, the COVID-19 pandemic halted study activities. The study was in the final 6 months of recruitment and intervention delivery. Enrollment and intervention delivery were transitioned to a virtual platform; process and adaptations are detailed elsewhere.⁴⁸

Patient-Centered Measures

The primary outcome measure was the reliable and valid Functional Assessment of Chronic Illness Therapy (FACIT) General scale that examines 4 domains comprising global QOL, including physical, social, emotional, and functional well-being subscales (score range, 0-108). Secondary patient-centered measures included the FACIT Spiritual Well-Being subscale and the Advance Care Planning Engagement Survey measuring 4 behavioral constructs (knowledge, contemplation, self-efficacy, and readiness) across a 5-point Likert scale, with higher numbers indicating greater engagement.⁴⁹ The Brief Pain Inventory⁵⁰ and the Edmonton Symptom Assessment Scale are evidence-based measures of pain and symptom severity, with higher scores corresponding to greater symptom burden.⁵¹ Participants also completed a process and satisfaction measure that was developed and used in our previous navigation studies.^35,36

Measures of Health Care Use

Blinded study team members (S.M.F., K.L., W.G., R.M.F.) conducted electronic health record (EHR) reviews to determine ACP documentation (MDPOA, study-specific AD, living will, and other comprehensive ADs [eg, Five Wishes]). Hospice use and length of stay and an established composite measure of aggressive care at the end of life (death in the hospital, no hospice use, or hospice of <3 days before death) were also assessed.^52,53 Date of death was confirmed using Colorado Vital Statistics records.

Data Collection

Navigators collected baseline measures from all patients following consent and prior to randomization. Blinded study personnel surveyed patients 3 months after enrollment to complete the patient-centered measures. At survey conclusion, the interviewer asked the patient if they had visits with the patient navigator. Those who received the navigator intervention completed additional questions focusing on intervention satisfaction. Six months after participant enrollment, the primary investigator (S.M.F.) and coinvestigators (K.L., W.G., R.M.F.) conducted EHR review for documented AD and at study month 46 to determine mortality and hospice outcomes.

Statistical Analysis

To assess the effectiveness of the randomization, we compared the 2 groups on participant-level sociodemographic variables. We assessed effect sizes (group imbalance defined as absolute value >0.20) and compared categorical variables using χ² tests (or Fisher exact tests) and continuous variables using t tests. To test the intervention’s effect on the outcome measures, t tests (or Wilcoxon tests if distribution was not normal) were used for change in continuous measures (QOL [as the primary outcome], FACIT Spiritual Well-Being, ACP engagement, pain and symptom experience, and hospice utilization in days for decedents), and χ² tests (or Fisher exact tests if cell size <5) were used for dichotomous outcome measures (hospice utilization, aggressive care at the end of life measures, adequacy of pain medication, and AD presence in the EHR). Satisfaction with the navigator intervention was described using frequencies and proportions. A sample of 186 patients (after 22.5% attrition from 240 enrolled) was estimated to have greater than 80% power in the primary outcome (QOL) to detect a medium effect size (Cohen f = 0.21) in change in QOL for a 2-sided test comparing 2 means at a type I error rate of α = .05. SAS, version 9.4 (SAS Institute), was used for all analyses. Statistical significance was defined as P < .05.

Results

Baseline Participant Characteristics

Overall, 593 patients were identified through the initial screening process. Of these, 197 were found to be ineligible. A study outreach letter was sent to the remaining 396 patients and reached 300. The navigators consented 209 patients, resulting in a 52.8% overall enrollment rate from eligible patients and 69.7% enrollment rate for those successfully contacted and invited to participate (Figure 1). There were no statistically significant differences between study groups across age, sex, or socioeconomic status (Table 1). Patients had a mean (SD) age of 63.6 (14.3) years. Among the 209 patients enrolled, 108 (51.7%) were male, 114 (55.1%) had less than a high school education, 162 (81.0%) reported an annual income less than $15 000, and 77 (36.8%) spoke Spanish as their primary language at home. Characteristics of patients able to complete follow-up measures were similar to those of all patients.

Figure 1. CONSORT Diagram.

Table 1. Participant Characteristics.

Characteristic	No. (%)a			Effect size	P valueb
	Total (N = 209)	Intervention (n = 104)	Control (n = 105)
Age, mean (SD), y	63.6 (14.3)	63.7 (13.5)	63.5 (15.1)	0.01	.94
Sex
Female	101 (48.3)	51 (49.0)	50 (47.6)	0.03	.84
Male	108 (51.7)	53 (51.0)	55 (52.4)	−0.03
Married or domestic partner	78 (37.7)	39 (37.9)	39 (37.5)	0.008	.96
Spanish primary language spoken at home	77 (36.8)	36 (34.6)	41 (39.0)	−0.09	.51
Born in the US	124 (60.2)	67 (65.0)	57 (55.3)	0.20	.15
<High school education	114 (55.1)	55 (53.9)	59 (56.2)	−0.05	.74
<$15 000 annual income	162 (81.0)	81 (81.0)	81 (81.0)	0.00	>.99
Not currently employed	202 (97.1)	103 (99.0)	99 (94.3)	0.35	.03c
Payment
Medicare Part A or Parts A and B	46 (22.4)	25 (24.8)	21 (20.2)	0.11	.97
Medicare Part C	3 (1.5)	1 (1.0)	2 (1.9)	−0.08
Medicare with private insurance	12 (5.9)	6 (5.9)	6 (5.8)	0.004
Medicaid	61 (29.8)	30 (29.7)	31 (29.8)	−0.002
Dually eligible	60 (29.3)	28 (27.7)	32 (30.8)	−0.07
Other pay source	23 (11.2)	11 (10.9)	12 (11.5)	−0.02
Diagnoses
Kidney disease	41 (19.8)	17 (16.5)	24 (23.1)	−0.17	.47
Congestive heart failure	53 (25.6)	27 (26.2)	26 (25.0)	0.03
Other cardiovascular disease	17 (8.2)	8 (7.8)	9 (8.7)	−0.03
Cirrhosis	24 (11.6)	15 (14.6)	9 (8.7)	0.18
All neurologic diagnoses	15 (7.2)	10 (9.7)	5 (4.8)	0.19
Pulmonary disease	32 (15.5)	13 (12.6)	19 (18.3)	−0.16
Other chronic disease	25 (12.1)	13 (12.6)	12 (11.5)	0.03
Multimorbidity	167 (80.7)	88 (85.4)	79 (76.0)	0.24	.08
Geographyd
Denver/Aurora	114 (54.5)	57 (54.8)	57 (54.3)	0.01	.97
Northern Colorado	59 (28.2)	28 (26.9)	31 (29.5)	−0.06
Southern Colorado	13 (6.2)	7 (6.7)	6 (5.7)	0.04
Mountain/frontier region	23 (11.0)	12 (11.5)	11 (10.5)	0.03
Setting/locatione
Large nonprofit health care systems	35 (16.7)	18 (17.3)	17 (16.2)	0.03	.99
Community health care clinics	23 (11.0)	12 (11.5)	11 (10.5)	0.03
Safety-net health care system/FQHC clinic	83 (39.7)	41 (39.4)	42 (40.0)	−0.01
Academic health care system	68 (32.5)	33 (31.7)	35 (33.3)	−0.03

Abbreviation: FQHC, federally qualified health center.

^a Percentages may not total 100% due to rounding. There were less than 5% missing data; excluding income for patient, there were less than 2% missing.

^b Intervention and control groups were compared using t tests for continuous variables (age), and χ² tests (or Fisher exact tests) for categorical variables (all other).

^c Fisher exact test.

^d Northern Colorado includes Greeley, Fort Collins, Loveland, 84th-Westminster, 144th, and Salud Family Health Centers; Southern Colorado includes Colorado Springs and Pueblo; and the mountain/frontier region includes Edwards, Avon, Eagle, Vail, Glenwood Springs, Basalt, Rifle, and Battlement Mesa.

^e Large nonprofit health care systems include Kaiser (Greeley, Fort Collins, Loveland, and Pueblo), and Centura (84th-Westminster, 144th, and Pueblo); community health care clinics include Mountain Family Health Center (Edwards, Glenwood Springs, and Basalt), Colorado Mountain Medical (Avon, Eagle, and Vail), and Grand River Health (Rifle and Battlement Mesa); safety-net health care system/FQHC clinic includes Denver Health (Denver) and Salud Family Health Centers in Northern Colorado; and academic health care system includes University of Colorado Health (Aurora, Fort Collins, Loveland, and Colorado Springs).

Intervention Delivery

Patients randomized to the intervention group were contacted by patient navigators to schedule home visits. Of the 104 patients receiving the intervention, 99 (95.2%) completed at least 1 home visit with the patent navigator, with 82 (78.8%) completing all 5 home visits. On average, patients in the intervention group received a mean (SD) of 4.5 (1.7) home visits; 14 patients requested additional visits. Average visits were a mean (SD) of 53 (18) minutes long. Fifty-three visits (11.8%) were conducted virtually during the COVID-19 pandemic. Participants in the intervention group reported high satisfaction with patient navigators (Figure 2).

Figure 2. Intervention Delivery (n = 448) and Patient Satisfaction (n = 81).

PN indicates patient navigator.

Primary Outcome

There were not statistically significant differences in total mean QOL scores, as measured by the FACIT General scale, nor the change in scores over time between the 2 groups. Total mean (SD) score in the control group was 65.0 (17.7) with a change of 4.3 (15.5) from baseline to follow-up, while in the intervention group participants had a mean (SD) score of 61.1 (17.6) with a change in score of 5.0 (16.5) (β, 0.8; 95% CI, −4.0 to 5.5; P = .75). Results from each of the individual subscales and related FACIT Spiritual Well-Being subscale are summarized in Table 2.

Table 2. Patient Outcomes.

Outcome	Mean (SD)a						Effect size	β (95% CI)	P valueb
	Intervention (n = 89; baseline values based on n = 104)			Control (n = 90; baseline values based on n = 105)
	Baseline	Follow-up	Change	Baseline	Follow-up	Change
FACIT General total score (range, 0-108)	61.1 (17.6)	65.5 (19.4)	5.0 (16.5)	65.0 (17.7)	68.6 (18.4)	4.3 (15.5)	0.04	0.8 (−4.0 to 5.5)	.75
Physical well-being subscale	14.7 (6.3)	15.7 (6.6)	0.9 (6.3)	15.4 (7.3)	15.9 (6.8)	0.7 (6.5)	0.03	0.2 (−1.7 to 2.1)	.84
Social/family well-being subscale	19.9 (6.3)	19.9 (6.4)	0.2 (6.8)	19.2 (5.6)	20.0 (6.4)	0.8 (6.3)	−0.1	−0.6 (−2.5 to 1.4)	.57
Emotional well-being subscale	13.8 (5.0)	14.7 (5.9)	1.0 (6.4)	15.1 (5.9)	16.2 (5.7)	1.3 (6.1)	−0.05	−0.3 (−2.2 to 1.5)	.72
Functional well-being subscale	13.0 (7.0)	15.5 (7.1)	2.7 (6.8)	15.4 (6.5)	16.7 (6.2)	1.3 (8.1)	0.2	1.4 (−0.8 to 3.6)	.21
FACIT Spiritual Well-Being score (range, 0-48)	34.1 (8.1)	35.9 (8.4)	2.5 (9.6)	35.8 (7.3)	35.7 (8.3)	−0.3 (9.6)	0.3	2.8 (−0.1 to 5.6)	.05
Advance Care Planning Engagement Survey score (range, 1-5)	2.6 (1.3)	3.4 (1.1)	0.8 (1.3)	2.7 (1.2)	2.9 (1.3)	0.1 (1.4)	0.5	0.7 (0.3 to 1.1)	<.001
Edmonton Symptom Assessment Scale (range, 0-100)	35.3 (21.6)	30.7 (20.6)	−4.9 (23.4)	28.6 (20.7)	28.6 (19.8)	−0.4 (20.6)	−0.2	−4.4 (−11.0 to 2.1)	.18
Brief Pain Inventory
Pain intensity score (range, 0-10)	4.8 (3.1)	4.2 (2.8)	−0.4 (2.6)	4.5 (2.8)	4.2 (2.7)	−0.5 (2.8)	0.04	0.1 (−0.7 to 0.9)	.79
Pain interference score (range, 0-10)	5.4 (3.5)	5.1 (3.3)	−0.2 (3.7)	5.2 (3.2)	5.0 (3.4)	−0.4 (3.7)	0.1	0.2 (−0.9 to 1.3)	.71
Stronger pain medication needed, No./total No. (%)	22/88 (25.0)			25/89 (28.1)			−0.1	−0.03 (−0.2 to 0.1)	.64
Needed more pain medication than prescribed, No./total No. (%)	18/87 (20.7)			18/89 (20.2)			0.01	0.005 (−0.1 to 0.1)	.94
Concerned about using too much pain medication, No./total No. (%)	6/86 (7.0)			10/89 (11.2)			−0.1	−0.04 (−0.1 to 0.04)c	.33
Had adverse effects from pain medication, No./total No. (%)	3/87 (3.4)			11/89 (12.4)			−0.3	0.09 (−0.2 to −0.01)c	.048d
Needed further information about pain medications, No./total No. (%)	5/88 (5.7)			5/89 (5.6)			0.004	0.0006 (−0.1 to 0.1)	.99

Abbreviation: FACIT, Functional Assessment of Chronic Illness Therapy.

^a There were less than 3% missing data.

^b Intervention and control groups were compared using t tests for continuous variables and χ² tests (or Fisher exact tests) for categorical variables.

^c Satterthwaite method.

^d Fisher exact test.

Secondary Outcomes

ACP

Patients randomized to the intervention group demonstrated a statistically significant larger improvement in ACP engagement compared with the control group (mean [SD], 0.8 [1.3] vs 0.1 [1.4]; β, 0.7; 95% CI, 0.3-1.1; P < .001). Patients in the intervention group were also more likely to have any type of AD documented in the EHR compared with patients in the control group (62 of 104 [59.6%] vs 28 of 105 [26.9%] β, 0.3; 95% CI, 0.2-0.5; P < .001). Additionally, 80 of the 104 patients (76.9%) in the intervention group vs 56 of the 105 patients (53.3%) in the control group had an MDPOA in the EHR (β, 0.2; 95% CI, 0.1-0.4; P < .001; Table 3). From a process perspective, patients in the intervention group, compared with those in the control group, were more likely to report that they had discussed preferences for their future health care with family (75 of 86 [87.2%] vs 58 of 89 [65.2%]; β, 0.2; 95% CI, 0.1-0.3; P < .001) and with their health care professional (63 of 86 [73.3%] vs 46 of 89 [51.7%]; β, 0.2; 95% CI, 0.1-0.4; P = .003).

Table 3. Advance Care Planning, Hospice Referrals, and End-of-Life Outcomes.

Outcome	No. (%)a		Effect size	β (95% CI)	P valueb
	Intervention (n = 104)	Control (n = 105)
Documented advance care planning in electronic health record	62 (59.6)	28 (26.9)	0.7	0.3 (0.2 to 0.5)	<.001
Documented health care decision-maker (medical durable power of attorney) in electronic health record	80 (76.9)	56 (53.3)	0.5	0.2 (0.1 to 0.4)	<.001
Hospice enrollment, No./total No. (%)	20/102 (19.6)	8/103 (7.8)	0.3	0.1 (0.02 to 0.2)c	.01
Patients deceasedd
Hospice referral	19/43 (44.2)	7 (21.2)	0.5	0.2 (0.01 to 0.4)	.04
Aggressive care at the end of lifee	27/42 (64.3)	28 (84.8)	−0.5	−0.2 (−0.4 to −0.003)	.046
Hospice length of stay, mean (SD), d	7.6 (17.7)	9.2 (46.7)	−0.05	−1.6 (−18.9 to 15.7)c	.85

^a There were less than 3% missing data.

^b Intervention and control groups were compared using Behrens-Fisher tests for continuous variables and χ² tests for categorical variables.

^c Satterthwaite method.

^d A total of 44 patients died in the intervention group and 33 in the control group.

^e Composite measure that included any of the following: no hospice, hospice fewer than 3 days, or death in the hospital.

Pain and Symptom Management

Both groups reported moderate (4-5 of 10) pain intensity (intervention mean [SD] pain score of 4.8 [3.1] with a change of −0.4 [2.6]; control mean [SD] pain score of 4.5 [2.8] with a change of −0.5 [2.8]; β, 0.1; 95% CI, −0.7 to 0.9; P = .79). There was not a statistically significant difference in mean (SD) pain interference scores between the 2 groups (intervention mean [SD] pain score of 5.4 [3.5] with a change of −0.2 [3.7]; control mean [SD] pain score of 5.2 [3.2] with a change of −0.4 [3.7]; β, 0.2; 95% CI, −0.9 to 1.3). Patients in both groups reported feeling very comfortable asking for pain medication if their pain was suboptimal. Eighteen of 89 patients (20.2%) in the control group reported that they did not need more pain medication than what was prescribed, similar to patients in the intervention group (18 of 87 patients [20.7%]; β, 0.005; 95% CI, −0.1 to 0.1; P = .94). Overall symptom scores, as measured by the Edmonton Symptom Assessment Scale, showed that mean (SD) symptom scores and changes were 35.5 (22.2) and −4.9 (23.4) in the intervention group and 29.1 (20.7) and −0.4 (20.6) in the control group (β, −4.4; 95% CI, −11.0 to 2.1; P = .18; Table 2).

Hospice and Health Care Utilization

Between the intervention and control groups, there were no statistically significant differences in patient-reported willingness to consider hospice for themselves (59 of 85 patients [69.4%] vs 57 of 89 patients [64.0%]; β, 0.1; 95% CI, −0.1 to 0.2; P = .45) or for a loved one (69 of 85 patients [81.2%] vs 65 of 89 patients [73.0%]; β, 0.1; 95% CI, 0.0 to 0.2; P = .20). Over the course of the study and follow-up period, 77 patients died. Of these patients, 19 of 43 (44.2%) and 7 of 33 (21.2%) decedents in the intervention and control groups, respectively, enrolled in hospice (β, 0.2; 95% CI, 0.01-0.4; P = .04). There were no statistically significant differences in mean (SD) length of stay between the intervention and control groups (7.6 [17.7] days vs 9.2 [46.7] days; β,−1.6; 95% CI, −18.9 to 15.7; P = .85). Of those who died, 27 of 42 patients (64.3%) in the intervention group scored positive on the composite measure of aggressiveness of care at the end of life compared with 28 of 33 patients (84.8%) in the control group (β, −0.2; 95% CI, −0.4 to 0.0; P = .046; Table 3).

Discussion

This study demonstrates that the patient navigator intervention did not improve the primary outcome, patient-reported QOL. However, patients receiving the navigator intervention improved on several key secondary outcomes, including ACP engagement and AD documentation. Participants reported that they were more likely to talk to family and health care professionals about their future health care decisions compared with those in the control group. Participants receiving the intervention vs control who died were more likely to be referred to hospice, although without statistically significant differences in length of stay. Relatedly, participants receiving the intervention were less likely to have very aggressive care at the end of life. Pain and symptom severity were similar across the 2 groups, demonstrating moderate pain and, overall, mild symptoms.

This article appreciably adds to the evidence base on the role and scope of the lay navigator in the care of persons experiencing serious, life-limiting illness. Lay navigators are not clinicians and do not have the training or professional scope to directly provide care for pain and symptom management or treat psychological symptoms associated with advanced illness. Their role is best described as preprimary PC, activating and empowering patients to seek primary PC from their clinicians. This approach was used in our previous trial of the lay navigator intervention for Hispanic persons diagnosed with advanced cancer, where patients in both groups reported high QOL and only mild pain.³⁶ The higher reported pain in the current study may be reflective of the many barriers that delay or impede a high-quality palliative approach for symptom management for noncancer diagnoses, including prognostic uncertainty, lack of effective management strategies for chronic nonmalignant pain, and health care professional lack of skills and knowledge.⁵⁴ Additionally, the PC interventions that have demonstrated the most substantial improvements in QOL have generally involved specialty PC with high-intensity clinical contacts.^53,55,56 It is, therefore, not surprising that a lay navigator who relies on a primary PC model is simply insufficient in intensity to affect QOL. While COVID-19 affected enrollment and we did not achieve our accrual goals, the follow-up rates were higher than estimated; therefore, we do not think that the negative findings for the primary outcome were due to lack of power. It is possible that the study was underpowered to detect improvements in spiritual well-being, which trended toward statistical significance.

In this trial, we confirmed the effectiveness of lay navigators to engage patients in ACP. We published similar outcomes in our cancer-focused lay navigator study.³⁶ Furthermore, trained navigators have been shown, in a large pragmatic Center for Medicare & Medicaid Innovation demonstration project, to improve ACP and to decrease end-of-life spending.⁵⁷ Despite the recent rhetoric over the lack of value of ACP, the value of focusing on preparing for future health care decisions remains high.^58,59,60 Not only does this process promote patient autonomy, but the conversations may also have important implications for caregiver anticipatory grief and bereavement.⁶¹ In this trial, patients receiving the intervention were more likely to have documented a decision-maker and complete an AD, a critical step in an interested-parties state, where decision-making does not immediately fall to next of kin, rather the interested parties of a patient convene to select a proxy decision-maker. Additionally, these patients were more likely to be willing to have conversations with both family and health care professionals about future health care decisions.

Patients in the intervention group had statistically significant higher hospice referral rates than those in the control group and above the national average, which is estimated between 38% and 40%.^62,63,64,65 In our prior randomized clinical trial of the navigator intervention for Hispanic persons with advanced cancer, we found very high hospice referral rates of 78% to 80%, possibly reflecting a statewide phenomenon of very high hospice uptake for persons with cancer that led to a ceiling effect or the possibility of contamination at the level of the oncology clinics.³⁶ In this study, where patients were recruited from diverse and more diffuse settings across large health care systems involving many health care professionals, we found statistically significant differences in hospice enrollment between the groups.

Limitations

Navigators conducted enrollment procedures, raising a potential concern of providing navigation prior to randomization. Additionally, tailored interventions inherently raise concerns of reproducibility and quality control. Navigators participated in extensive training and role-playing prior to intervention implementation. We provided navigators with standardized responses about how to contact a clinic social worker or nurse in the event that patients or caregivers, prior to randomization, asked for assistance. We also used careful tracking and documentation of navigator contacts to address these concerns.^66,67

Another potential limitation is the fact that this study took place in Colorado, where the Hispanic population is largely of Mexican origin. Further testing of this lay navigator intervention in other diverse Hispanic communities is warranted. However, the inclusion of academic, safety-net, and community primary care and specialty clinics represents an enormous strength and offers a level of external validity that is rarely seen in PC research.

Conclusions

In this randomized clinical trial, a culturally tailored patient navigator intervention did not improve QOL. However, the intervention was highly valued by patients and demonstrated improvement in ACP outcomes, hospice utilization, and aggressiveness of care at the end of life.

Supplement 1.

Trial Protocol

Supplement 2.

Data Sharing Statement