Effect of the Serious Illness Care Program on Health Care Utilization at the End of Life for Patients with Cancer

Abstract

Objectives:

To determine the effect of the Serious Illness Care Program on health care utilization at the end of life in oncology.

Design:

Analysis of the secondary outcome of health care utilization as part of a cluster-randomized clinical trial that ran from 2012 to 2016. Clinicians in the intervention group received training, coaching, and system supports to have discussions with patients using a Serious Illness Conversation Guide (SICG); clinicians in the control arm followed usual care.

Setting/Subject:

Patients with advanced cancer who died within two years of enrollment at the Dana-Farber Cancer Institute.

Measurement:

Health care utilization was abstracted from the electronic medical record using the National Quality Forum (NQF)-endorsed indicators of aggressive cancer care at the end of life and scored from 0 to 6 (one point for each aggressive indicator); t tests and chi-square tests were used to determine differences between intervention and control patients.

Results:

The charts of 159 patients who died were reviewed. Neither the main outcome of mean number of aggressive indicators (0.9 vs. 0.9, p = 0.84) nor the proportion of patients with any aggressive care (49% intervention [95% CI: 40–57] vs. 54% control [95% CI: 42–67]) differed between patients in the intervention and control groups.

Conclusion:

In this analysis of a secondary outcome from a randomized clinical trial of the Serious Illness Care Program, intervention and control patients had similar end-of-life health care utilization as measured by the mean number of NQF-endorsed indicators. Future research efforts should focus on studying the strategies by which communication about patients' prognosis, values, and goals leads to personalized care plans.

Introduction

Healthcare utilization that is aligned with patients' priorities has been identified as a key outcome of high-quality person-centered serious illness care.^1–3 Prior observational studies have demonstrated that conversations about patients' end-of-life (EOL) goals are associated with more and earlier hospice use, fewer hospitalizations, and improvements in goal-concordant care, satisfaction, and quality of life.^4–6 The National Quality Forum (NQF) and the American Society of Clinical Oncology have jointly endorsed indicators of aggressive care that reflect poor quality cancer care at the EOL, including overuse of chemotherapy, underuse of hospice services, and misuse of treatments that lead to hospitalizations, emergency department (ED) visits, and intensive care unit (ICU) stays.^7,8 Conjoining improved goals-based communication and goal-aligned care with more appropriate health care utilization is a sought-after outcome for health systems and patients.

Randomized trials of palliative care interventions, with a strong focus on goals-based communication,⁹ have demonstrated improvements in patient outcomes^10,11 and reductions in aggressiveness of care.¹¹ Studies of interventions designed to improve communication outside of specialty palliative care, however, have demonstrated mixed results on patient outcomes and health care utilization.^12–17 We have previously reported primary and secondary outcome findings from a randomized clinical trial of a communication quality improvement intervention, the Serious Illness Care Program.^18,19 In this study, we report the secondary outcome of health care utilization from this trial using the NQF-endorsed EOL care indicators.⁸

Methods

Study design

This is an analysis of secondary health care utilization outcomes in a cluster-randomized trial at the Dana-Farber Cancer Institute (DFCI) (2012–2016).²⁰ Clinicians were randomized in clusters (units of clinicians within a disease center that typically included one advanced practice clinician and two or three physicians) at enrollment to intervention or control. The study was approved by the DFCI Institutional Review Board.

Participants and eligibility criteria

Physicians and advanced practice clinicians from the DFCI were eligible to participate. To identify eligible patients (who were receiving oncology care at DFCI and were at least 18 years old), all clinicians used the surprise question: Would you be surprised if this patient died in the next year?²¹ Only patients for whom the clinician responded “no, I would not be surprised” were eligible. Patients who had cognitive impairment, were not English-speaking, or could not identify a caregiver to participate were excluded. Patients who died within two years of enrollment were included in this analysis.

Sample size

We powered the study based on the co-primary outcomes, goal-concordant care and peacefulness at EOL, reported separately.^18,20

Intervention

Clinicians in the intervention group received communication skills training, coaching, and system supports (e.g., e-mail reminder, electronic health record documentation template) to have discussions with their patients using a Serious Illness Conversation Guide. Intervention patients received supportive materials (e.g., patient preparation letter). Clinicians received $150 gift cards for participation; patients did not receive remuneration for participation. Additional details of the intervention are described elsewhere.²⁰

Usual care

Patients of clinicians enrolled in the control group received usual oncology care. Enrolled control clinicians and patients did not receive any of the aforementioned intervention components.

Blinding

Patient participants were blinded to study arm. Clinicians were not blinded. The data abstractor (L.K.) was blinded to the randomization assignment.

Demographic characteristics

Clinicians and patients in both arms completed a survey at baseline to collect demographic characteristics.

Outcomes

We followed all patients until death or for up to 24 months. Based on the NQF-endorsed indicators, we collected the occurrence and dates of the following from the electronic medical record for patients who died: (1) any chemotherapy in the last 14 days; (2) ≥2 hospitalizations in the last 30 days; (3) ≥2 ED visits in the last 30 days; (4) ≥1 ICU stay in the last 30 days; (5) no hospice use or hospice use less than three days; (6) death in acute care hospital. We also recorded receipt of inpatient or outpatient palliative care visits. Our a priori defined outcome was the mean number of indicators per patient (0–6).¹²

Statistical analysis

Baseline characteristics for deceased patients are presented as means with 95% confidence intervals or as a count and percent. With an a priori observed standard deviation of the number of indicators equal to 1.17, we had 80% power to detect a 0.58 difference in mean number of indicators (score 0–6) between the intervention and control arms, using a robust t test (not assuming normality) and clustering by clinician team,²² with observed Intraclass Correlation Coefficient (ICC) = 0.019.²³ For additional outcomes, a robust t test accounting for clustering was used for continuous outcomes and a chi-square test accounting for clustering was used for all proportions. A p value <0.05 was considered statistically significant. Analyses were conducted with intention-to-treat. All statistical analyses were performed using SAS software, version 9.4 (SAS Institute, Inc.).

Results

A total of 91 oncology clinicians enrolled in the trial; 379 patients enrolled, and 278 patients had analyzable data (of whom 58% died during the study period (n = 161)).^18,19 We reviewed the charts of 157 of the 161 patients; we excluded four patients who died within 30 days of enrollment. Clinician characteristics did not differ between arms.¹⁹ Baseline characteristics of patients were similar (Table 1).

Table 1.

Patient Characteristics

	Intervention n = 74	Control n = 83
Age in years, mean (95% CI)	62 (57–67)	62 (57–67)
Female gender, n (%)	41 (55)	45 (54)
Race, n (%)^a
White	68 (92)	73 (88)
Black or African American	0 (0)	1 (1.2)
Other	5 (6.8)	2 (2.4)
Ethnicity Hispanic, n (%)^a	2 (2.7)	0 (0)
Married/partnered, n (%)	61 (82)	66 (80)
Income ≥$75,000, n (%)	40 (54)	35 (42)
Disease center, n (%)^b
Breast oncology	20 (27)	25 (30)
Gastrointestinal, genitourinary, head and neck, neurology, sarcoma, thoracic, other	50 (68)	52 (63)
Hematologic malignancies, lymphoma	4 (5.4)	6 (7.2)
Health insurance type, n (%)
Medicare	39 (53)	39 (47)
Medicaid/mass health	3 (4.1)	3 (3.6)
Private	31 (42)	36 (43)
No insurance	0 (0)	1 (1.2)
Patient-reported health status, n (%)
Relatively healthy and not seriously ill	8 (11)	5 (6.0)
Relatively healthy and terminally ill	45 (61)	51 (61)
Seriously but not terminally ill	14 (19)	15 (18)
Seriously and terminally ill	6 (8.1)	8 (9.6)
College, graduate, or professional school, n (%)	61 (82)	67 (81)

Percentages do not sum to 100 owing to rounding. Because the percentage missing for any variable was <7%, missing data are not reported in this table. Calculations for percentages were based on nonmissing data.

Race or ethnic group was self-reported.

Disease center did not include gynecologic oncology because of a concurrent trial being conducted at that center.

Intervention and control patients had similar EOL health care utilization (Table 2). The mean number of aggressive care indicators (0.9 [95% CI: 0.6–1.1] vs. 0.9 [95% CI: 0.6–1.2], p = 0.84) did not differ between arms, nor did the proportion of patients with any aggressive care indicator (49% intervention [95% CI: 40–57] vs. 54% control [95% CI: 42–67]). Receipt of palliative care between groups was similar (30% intervention [95% CI: 21–38] vs. 34% control [95% CI: 22–46]). Table 2 includes additional data by indicator.

Table 2.

Aggressiveness of Care Indicators

	Intervention n = 74			Control n = 83			p
Main outcome
Number of aggressive indicators, mean (95% CI)	0.9		0.6–1.1	0.9		0.6–1.2	0.84
Secondary outcomes	n	%	95% CI	n	%	95% CI
Receipt of any aggressive care, n (%) (95% CI)	36	49	40–57	45	54	42–67
Chemotherapy administration
Any chemotherapy within 14 days of death, n (%) (95% CI)	5	6.8	3.0–11	9	11	5.1–17
Hospitalization
≥2 hospitalizations in last 30 days, n (%) (95% CI)	5	6.8	2.5–11	6	7.2	2.3–12
ED visits
≥2 ED visits in last 30 days, n (%) (95% CI)	0	0.0	NA	3	3.6	0.0–7.5
No hospice services or <3 days	32	43	35–52	38	46	30–62
No admission to hospice, n (%) (95% CI)	29	39	29–49	34	41	27–55
Admission to hospice <3 days before death, n (%) (95% CI)	3	4.1	0.7–7.4	4	4.8	0.0–9.7
ICU
Any ICU admission within 30 days of death, n (%) (95% CI)	9	12	3.1–21	4	4.8	0.5–9.1
Location of death
Death in acute care setting, n (%) (95% CI)	12	16	5.3–27	14	17	7.7–26

ED, emergency department; ICU, intensive care unit; N/A, not applicable.

Discussion

In this analysis of a secondary outcome from a randomized clinical of the Serious Illness Care Program, we found that intervention and control patients had similar EOL health care utilization as measured by the mean number of NQF-endorsed indicators.

It stands to reason that if many patients value comfort-focused care, then communication interventions designed to improve goals-based communication could result in changes in health care utilization. There are several potential interpretations of the null findings. First, perhaps the intervention did not target the right mechanisms, or was not sufficiently potent, to translate early serious illness communication into changes in health care utilization at the end of life. The trial was designed to facilitate a single conversation about patients' values and goals earlier in the illness course.¹⁹ It did not include training to translate conversations into action²⁴ or a prompt for a follow-up conversation later in the illness course. In addition, nurses and social workers, key partners in serious illness communication, were not included in this study; an interprofessional approach may be central to translating conversations into clinical action.^25,26

Second, our power was only sufficient to detect a very large effect size in receipt of any aggressive care (60% change) at the EOL, which is unlikely given that prior randomized and observational studies have demonstrated on average 8%–20% differences.^4,11 Although observational studies in oncology have shown associations between EOL conversations and lower rates of aggressive care, oncology patients in these studies may be more likely to have conversations because they want less aggressive care or are less anxious about the future and better able to develop adaptive coping strategies.^4,27,28 In our randomized study, it may be that patients who did not want to have conversations did not enroll. Indeed, the proportion of patients with conversations in our control group was high (96% intervention vs. 79% control),¹⁹ which makes it more difficult to demonstrate changes.

Lastly, the validity of different constructs for measuring “aggressive care” at EOL remains unknown. In the Temel et al. randomized study of specialty palliative care,¹¹ which demonstrated a significant reduction in aggressive care, the study team defined aggressive care as a patient who received any one of the following: chemotherapy in the last 14 days, no hospice, or fewer than three days of hospice. We used a different construct in our study—the mean number of all six health care utilization indicators identified by Earle et al., which included hospitalizations, ICU, and ED visits.⁸ In addition, only 11% of control patients in our study received chemotherapy in the last 14 days of life, which is a low proportion that may suggest a potential floor effect. Measurements of EOL health care utilization at the population level must be interpreted cautiously, as we do not know patients' individual goals and preferences. We need more research to understand how serious illness communication interventions lead to certain outcomes, especially care that aligns with patient goals.

Limitations

This study has several limitations in addition to being underpowered to detect nonrobust differences. Electronic medical record chart abstraction may underestimate utilization because it only captures care within Partners' Health Care facilities. The trial took place at a large specialized cancer center with a homogenous patient population and may not be generalizable to other settings.

Conclusion

The Serious Illness Care Program communication intervention did not demonstrate a change in health care utilization at the EOL in a specialty oncology center, as measured by the NQF-endorsed EOL care indicators for cancer care. However, limitations in the power for this analysis weaken the finding's strength and the inferences that can be drawn. Future research efforts should focus on more sensitive and patient-informed measurements of health care utilization as well as studying the mechanisms by which communication about patients' prognosis, values, and goals leads to care that align with what matters most to patients.

Funding Information

Supported by the John A. Hartford Foundation. The funder did not play a role in the design, execution, or writing of this work.

Footnotes

Author Disclosure Statement

Dr. Block receives compensation as Palliative Care Editor from Up to Date.

References

Committee on Approaching Death: Addressing Key End of Life

Issues

, Institute of Medicine: Dying in America: Improving Quality and Honoring Individual Preferences Near the End of Life. Washington, DC: National Academies Press, 2015.

Sanders

, Curtis

, Tulsky

: Achieving goal-concordant care: A conceptual model and approach to measuring serious illness communication and its impact. J Palliat Med, 2018; 21(S2):S17–S27.

Teno

, Montgomery

, Valuck

, et al.: Accountability for community-based programs for the seriously Ill. J Palliat Med, 2018; 21(S2):S81–S87.

Wright

, Zhang

, Ray

, et al.: Associations between end-of-life discussions, patient mental health, medical care near death, and caregiver bereavement adjustment. JAMA, 2008; 300:1665–1673.

Heyland

, Allan

, Rocker

, et al.: Discussing prognosis with patients and their families near the end of life: Impact on satisfaction with end-of-life care. Open Med, 2009; 3:e101–e110.

Leung

, Udris

, Uman

, Au

: The effect of end-of-life discussions on perceived quality of care and health status among patients with COPD. Chest, 2012; 142:128–133.

Earle

, Park

, Lai

, et al.: Identifying potential indicators of the quality of end-of-life cancer care from administrative data. J Clin Oncol, 2003; 21:1133–1138.

Earle

, Landrum

, Souza

, et al.: Aggressiveness of cancer care near the end of life: Is it a quality-of-care issue?. J Clin Oncol, 2008; 26:3860–3866.

Yoong

, Park

, Greer

, et al.: Early palliative care in advanced lung cancer: A qualitative study. JAMA Intern Med, 2013; 173:283–290.

10.

Zimmermann

, Swami

, Krzyzanowska

, et al.: Early palliative care for patients with advanced cancer: A cluster-randomised controlled trial. Lancet, 2014; 383:1721–1730.

11.

Temel

, Greer

, Muzikansky

, et al.: Early palliative care for patients with metastatic non-small-cell lung cancer. N Engl J Med, 2010; 363:733–742.

12.

Epstein

, Duberstein

, Fenton

, et al.: Effect of a patient-centered communication intervention on oncologist-patient communication, quality of life, and health care utilization in advanced cancer: The VOICE Randomized Clinical Trial. JAMA Oncol, 2017; 3:92–100.

13.

Curtis

, Back

, Ford

, et al.: Effect of communication skills training for residents and nurse practitioners on quality of communication with patients with serious illness: A randomized trial. JAMA, 2013; 310:2271–2281.

14.

Tulsky

, Arnold

, Alexander

, et al.: Enhancing communication between oncologists and patients with a computer-based training program: A randomized trial. Ann Intern Med, 2011; 155:593–601.

15.

Clayton

, Butow

, Tattersall

MHN

, et al.: Randomized controlled trial of a prompt list to help advanced cancer patients and their caregivers to ask questions about prognosis and end-of-life care. J Clin Oncol, 2007; 25:715–723.

16.

Connors

: A controlled trial to improve care for seriously III hospitalized patients. JAMA, 1995; 274:1591.

17.

Patel

, Sundaram

, Desai

, et al.: Effect of a lay health worker intervention on goals-of-care documentation and on health care use, costs, and satisfaction among patients with cancer: A randomized clinical trial. JAMA Oncol, 2018; 4:1359–1366.

18.

Bernacki

, Paladino

, Neville

, et al.: Effect of the Serious Illness Care Program in outpatient oncology: A cluster randomized clinical trial. JAMA Intern Med, 2019; 179:751–759.

19.

Paladino

, Bernacki

, Neville

, et al.: Evaluating an intervention to improve communication between oncology clinicians and patients with life-limiting cancer: A cluster randomized clinical trial of the Serious Illness Care Program. JAMA Oncol, 2019; 5:801–809.

20.

Bernacki

, Hutchings

, Vick

, et al.: Development of the Serious Illness Care Program: A randomised controlled trial of a palliative care communication intervention. BMJ Open, 2015; 5:e009032.

21.

Moss

, Lunney

, Culp

, et al.: Prognostic significance of the “surprise” question in cancer patients. J Palliat Med, 2010; 13:837–840.

22.

Ware

, Lipsitz

, Speizer

: Issues in the analysis of repeated categorical outcomes. Stat Med, 1988; 7:95–107.

23.

Donner

, Klar

: Cluster randomisation trials in epidemiology: Theory and application. J Stat Plan Inference, 1994; 42:37–56.

24.

Geerse

, Lamas

, Sanders

, et al.: A qualitative study of serious illness conversations in patients with advanced cancer. J Palliat Med, 2019; 22:773–781.

25.

Lakin

, Koritsanszky

, Cunningham

, et al.: A systematic intervention to improve serious illness communication in primary care. Health Aff (Millwood), 2017; 36:1258–1264.

26.

Lakin

, Benotti

, Paladino

, et al.: Interprofessional work in serious illness communication in primary care: A qualitative study. J Palliat Med, 2019; 22:751–763.

27.

Mack

, Cronin

, Keating

, et al.: Associations between end-of-life discussion characteristics and care received near death: A prospective cohort study. J Clin Oncol, 2012; 30:4387–4395.

28.

Nipp

, Greer

, El-Jawahri

, et al.: Coping and prognostic awareness in patients with advanced cancer. J Clin Oncol, 2017; 35:2551–2557.