Early Palliative Care Reduces End‐of‐Life Intensive Care Unit (ICU) Use but Not ICU Course in Patients with Advanced Cancer

Early palliative care significantly reduced intensive care unit (ICU) use at the end of life but did not change ICU events. This study supports early initiation of palliative care for advanced cancer patients before hospitalizations and intensive care.

Abstract

Background.

Early palliative care for advanced cancer patients improves quality of life and survival, but less is known about its effect on intensive care unit (ICU) use at the end of life. This analysis assessed the effect of a comprehensive early palliative care program on ICU use and other outcomes among patients with advanced cancer.

Patients and Methods.

A retrospective cohort of patients with advanced cancer enrolled in an early palliative care program (n = 275) was compared with a concurrent control group of patients receiving standard care (n = 195) during the same time period by using multivariable logistic regression analysis. The multidisciplinary outpatient palliative care program used early end‐of‐life care planning, weekly interdisciplinary meetings to discuss patient status, and patient‐reported outcomes assessment integrated within the electronic health record.

Results.

Patients in the control group had statistically significantly higher likelihood of ICU admission at the end of life (odds ratios [ORs]: last 6 months, 3.07; last month, 3.59; terminal admission, 4.69), higher likelihood of death in the hospital (OR, 4.14) or ICU (OR, 5.57), and lower likelihood of hospice enrollment (OR, 0.13). Use of chemotherapy or radiation did not significantly differ between groups, nor did length of ICU stay, code status, ICU procedures (other than cardiopulmonary resuscitation), disposition location, and outcomes after ICU admission.

Conclusion.

Early palliative care significantly reduced ICU use at the end of life but did not change ICU events. This study supports early initiation of palliative care for advanced cancer patients before hospitalizations and intensive care.

Implications for Practice.

Palliative care has shown clear benefit in quality of life and survival in advanced cancer patients, but less is known about its effect on intensive care. This retrospective cohort study at a university hospital showed that in the last 6 months of life, palliative care significantly reduced intensive care unit (ICU) and hospital admissions, reduced deaths in the hospital, and increased hospice enrollment. It did not, however, change patients’ experiences within the ICU, such as number of procedures, code status, length of stay, or disposition. The findings further support that palliative care exerts its benefit before, rather than during, the ICU setting.

Introduction

The benefits of early specialty palliative care in the setting of advanced cancer have been reported in several recent publications. Recipients have higher satisfaction with care and fewer emergency room, hospital, and intensive care unit (ICU) admissions; they have better mood and quality of life, experience less depression, and are more likely to die at home rather than in the hospital [1], [2], [3], [4], [5], [6]. In one prospective study, despite the receipt of less aggressive care with lower cost, a significant survival advantage was shown in patients with metastatic non‐small cell lung cancer who received early palliative care compared with those who did not [7].

Currently, one in five deaths in the United States use ICU services, compared with one in three deaths at home [8]. ICU use in the last month of life is increasing, as are costs of ICU care as a whole [9], [10], [11]. This is especially prominent in the United States, where the ratios of ICU bed‐to‐population (20/100,000) and ICU‐to‐hospital bed (9/100) are the highest in the world [12]. Eighty percent of all costs for patients who die in the hospital are generated from those that included an ICU stay [8].

The care provided to patients with advanced cancer at the end of life is widely variable and aggressive. Divergent patient views on end of life abound, although many patients prefer comfort over intensive care and would prefer to die in a home‐like setting rather than a hospital‐like setting [13], [14], [15]. Additionally, some ICU interventions may be of limited benefit to advanced cancer patients dying of end‐stage organ failure, for example, intubation of patients with progressive lung cancer and poor performance status [16], [17]. Still, many patients are reluctant or ill equipped to make decisions that might forego the most aggressive interventions [18]. Recent prospective and systematic review data show trends toward decreased number of hospitalizations, shorter length of stay (LOS), and fewer ICU admissions in patients receiving palliative care interventions [1], [15], [19], [20]. Given the potential for improved patient outcomes, better satisfaction with care, avoidance of undesired or ineffective interventions, and cost savings, there are calls for improved advanced care planning and communication, as well as earlier integration of palliative care, for cancer patients [21], [22], [23]. This study presents results from a multidisciplinary outpatient palliative care program designed to improve the quality of life of patients with incurable malignancies through early integration of palliative care.

Materials and Methods

The CARE Track (Comprehensive Assessment with Rapid Evaluation and Treatment) program is a multiyear study of a phased palliative care intervention implemented by the University of Virginia Emily Couric Clinical Cancer Center that began in 2012. Early results from this same program, not related to ICU admission, have been recently published [6]. The goal of the program is to improve the quality of life of patients with incurable malignancies while reducing costs and decreasing hospitalization at the end of life.

The Palliative Care Service at the University of Virginia includes an outpatient clinic housed in the Cancer Center, inpatient consultation, a palliative care unit, and home hospice components. All are staffed by the same physicians and nurse practitioners, ensuring continuity of care across settings. Patients were identified at tumor boards, upon hospital admission, and by direct referral at the discretion of providers. Patients referred to the intervention were followed concurrently with medical oncology, radiation oncology, and surgical oncology for symptom management and end‐of‐life care planning, and they received early care planning discussions with palliative care specialists. There was no protocol‐specified time window from diagnosis to CARE Track referral, but great efforts were made to identify patients as early in their cancer course as possible. In year 1, the program implemented a weekly interdisciplinary meeting (Supportive Care Tumor Board) to coordinate care and identify patients in the program with worsening symptoms, with social situations requiring support services, and with imminent transitions of care. Beginning in year 2, patients completed a patient‐reported outcomes (PRO) assessment that included a series of health domains measured by using the National Institutes of Health Patient‐Reported Outcomes Measurement Information System and other symptom‐specific assessment items. Results from the PRO assessments were made available from within the electronic health record as a guide to symptom management and decision‐making. All patients participating in the CARE Track program who were deceased as of February 2015 are included in the analysis. Patients with stage IV solid tumors or other advanced cancer identified as incurable by the referring provider were referred to the CARE Track program at the discretion of their oncologist.

A control group of patients was identified retrospectively to assess differences in outcomes between patients participating in the intervention and patients receiving standard care. The control group was identified from among all patients not enrolled in the outpatient CARE Track program who received care within the cancer center during the same period, who had similar cancer diagnoses, and who died during the same period. Patients in the control group were identified through analysis of electronic health record data included in the Clinical Data Repository, a data warehouse managed by the Clinical Informatics Division of the Department of Public Health Sciences that contains a comprehensive collection of inpatient and outpatient histories for care provided at the University of Virginia Health System [24]. The control patient group included individuals who received palliative care and/or hospice planning services at the very end of life who were not enrolled in the CARE Track program.

Patients in both the CARE Track and control groups were classified with regard to sex, self‐reported race, marital status, religion, age at death, total number of Charlson comorbidities, stage, cancer diagnosis, presence of metastatic disease, and stem cell transplant recipient status. All data were obtained directly from individual chart review. The occurrence of the following 10 outcomes was also assessed for each patient in the study population: ICU admission in last 6 months of life, ICU admission in last month of life, ICU stay during terminal admission, death in hospital, death in ICU, enrollment in hospice, receipt of chemotherapy within 6 months of death, receipt of chemotherapy within 2 weeks of death, receipt of radiation within 6 months of death, and receipt of radiation within 2 weeks of death. Secondary outcomes for patients admitted to an ICU were also collected, including procedures, code status on ICU admission and discharge, disposition location, ICU LOS, and timing of ICU admission. This project was reviewed and approved by the University of Virginia Institutional Review Board.

Multivariable logistic regression analysis was used to estimate the difference in the odds of each outcome event for patients in the CARE Track compared with control patient groups, for each of the 10 outcomes. For each outcome, the difference in the odds of the event was adjusted for concurrent differences among patients in sex, race, marital status, religion, age at death, total Charlson comorbidity score, stage, cancer diagnosis, and presence of metastatic disease. Each model was specified a priori with regard to the 10 outcomes assessed and 10 covariates included in each model. Because early referral to palliative care was the intervention in this project, the time to referral was felt to be confounding for multivariate modeling and thus was not included.

Differences in the proportional distribution of demographic and clinical characteristics between the CARE Track and control group patients were assessed by using the Wald chi‐square test statistic for categorical variables and using the t test statistic for differences in means. The Wald chi‐square test statistic was also used to assess the statistical significance of the difference in the adjusted odds of each primary outcome event between patients in each group. The significance of differences in the adjusted odds of each primary outcome event between patients in the study groups was expressed in terms of 95% confidence intervals (CIs).

The relative contribution of each covariate to the overall predictive performance of each model was assessed by calculating type III analysis effects (Wald test statistics) for each model covariate to account for differences in covariate degrees of freedom. The relative magnitudes of the type III Wald test statistics reflect the proportion of the total model log‐likelihood independently explained by each model covariate. The capacity of each model to discriminate between patients with and without the predicted event was measured by using the C statistic [25], [26]. A C statistic value of 0.5 indicates that the model provides no predictive discrimination, whereas a value of 1.0 indicates perfect discrimination. A p value <.001 was used as the threshold standard for statistical significance for all comparisons. All calculations were performed using the Statistical Analysis System (SAS) software, version 9.4, SAS Institute Inc., Cary, NC, USA. http://www.sas.com/en_us/home.html

Results

Two hundred seventy‐five deceased CARE Track program patients were identified for analysis. Of these, 195 were identified for inclusion in the control group (Table 1). Overall, both groups were composed predominantly of white, married, Christian patients with stage IV/metastatic disease. The CARE Track group included a larger proportion of women (59% vs. 41%; p < .001), had a younger mean age at death (59.7 years vs. 65.9 years; p < .001), had fewer comorbidities (Charlson score, 8.9 vs. 11.1; p < .001), had fewer stage IV patients (61% vs. 83%; p < .001), and had fewer patients with metastatic disease (67% vs. 89%; p < .001). The CARE Track group included more patients with head and neck cancer (10.9% vs. 3.4%) and breast cancer (8.0% vs. 3.4%) and fewer non‐small cell lung cancer cases (11.6% vs. 26.7%; p < .001; complete data available in supplemental online Table 1).

Table 1. Patient characteristics.

Unless otherwise noted, values are presented as number (percentage) of patients. Values expressed with a plus/minus sign are the mean ± SD.

Abbreviations: CARE Track, Comprehensive Assessment with Rapid Evaluation and Treatment; NSCLC, non‐small cell lung cancer; SCLC, small cell lung cancer.

Compared with the CARE Track group, the control group had significantly higher adjusted odds of ICU admission during the last 6 months of life (odds ratio [OR], 3.1; 95% CI, 1.81–5.21) (Fig. 1; Table 2), during the last month of life (OR, 3.6; 95% CI, 1.96−6.59), during the terminal admission (OR, 4.7; 95% CI, 2.27–9.72), and higher odds of death in the hospital (OR, 4.1; 95% CI, 2.42–7.08) or death in the ICU (OR, 5.6; 95% CI, 1.98–15.69). Patients in the control group were also significantly less likely to be enrolled in hospice (OR, 0.13; 95% CI, 0.06–0.26) (Fig. 1).

Figure 1.

Odds ratios for primary outcomes. Multivariable adjusted odds ratios obtained for the intervention covariate, for each of the 10 outcomes listed along the vertical axis of the plot, with 95% confidence intervals depicted for each odds ratio estimate. Each of the plotted odds ratios represents the likelihood of the corresponding event among patients in the standard care only control group, compared to patients in the CARE Track intervention group. Each odds ratio is adjusted for differences between the two groups in sex, race, marital status, religion, age at death, total number of Charlson comorbidities, stage, cancer diagnosis, and presence of metastatic disease.

Abbreviations: CARE Track, Comprehensive Assessment with Rapid Evaluation and Treatment; ICU, intensive care unit.

Table 2. Primary outcomes.

Unless otherwise noted, values are the number (percentage) of patients.

Abbreviation: CARE Track, Comprehensive Assessment with Rapid Evaluation and Treatment; ICU, intensive care unit.

Comparison of the type III Wald test statistics of the covariates included in each model demonstrates that the CARE Track versus control group covariate provided the most significant contribution to the prediction of ICU admission at each time endpoint, death in the hospital, death in ICU, and hospice enrollment (supplemental online Fig. 1). Each statistical model demonstrated moderate to good discrimination between patients with and without the predicted event, with C statistics ranging between 0.71 and 0.86. Complete results for each of the 10 multivariable logistic regression models are available in the supplemental online data (supplemental online Tables 5–24).

There were no statistically significant differences between CARE Track and control groups in individual or total ICU procedures (supplemental online Tables 2–4), other than a reduction in cardiopulmonary resuscitation (CPR) in the CARE Track group. CARE Track and control group patients were similar with regard to code status on admission to and discharge from the ICU, disposition location, and chemotherapy received after ICU admission. There were also no significant differences in ICU LOS, time from cancer diagnosis to ICU admission, time for ICU discharge to death, and time from last chemotherapy to death.

Discussion

Cancer is a major cause of death in the United States, with nearly 600,000 deaths estimated in 2015 [27]. Given the burden of disease, high mortality and morbidity, and high inpatient costs incurred at the end of life, there is great interest in improving outpatient management for these patients in order to reduce hospitalizations and ICU admissions where appropriate [22], [23]. Patients with advanced cancer frequently have an illness trajectory that may be relatively stable before a rapid decline in performance status and increase in symptom burden in the last 6 months of life, as disease progression and iatrogenic toxicities mount [28], [29]. This is frequently accompanied by an increase in hospital admissions, decreased quality of life, increased costs, and declining patient satisfaction [30]. Hospitals, especially ICUs, are not where patients prefer to die, yet ICU admissions at the end of life are increasing, probably because of patients’ and providers’ lack of comfort with a more palliative approach [11], [18]. Pioneering work in patients with metastatic lung cancer has demonstrated that early integration of palliative care extends survival and addresses the “triple aim” of better health, improved care, and lower costs [7], [23], [31]. Our study addresses this issue by looking in detail at patients with advanced cancer at the end of life and their experience in the ICU, specifically focusing on whether early integration of palliative care influences key patient outcomes.

Data presented here indicate that early integration of palliative care can substantially reduce ICU admissions near the end of life, including an absolute reduction of nearly 50% during the period 6 months before death (Fig. 1; Table 2). Large reductions are also demonstrated for the proportion of patients who die in the hospital and who die in the ICU. Enrollment in hospice was also substantially increased. Our results complement the results of several prior studies that have also demonstrated improved patient outcomes associated with early integration of palliative care [5], [19]. There is no consensus on the definition of early referral, but prior studies have used 8 weeks or 30–60 days [5], [7]. Previous studies have shown a reduction in chemotherapy use at the end of life for some patients [32], but this was not shown in this study (p = .07).

Secondary outcome analysis demonstrates that patients had similar experience once admitted to the ICU, regardless of whether they were receiving palliative care (supplemental online Tables 2–4). Other than reduced CPR in the CARE Track group, there were no differences in ICU procedures, code status, disposition, or timing of ICU admission during their illness. This equivalency in ICU procedures indicates similar experiences in the ICU for patients with advanced cancer, regardless of palliative care receipt. In other words, the involvement of a palliative care specialist did not predispose ICU practitioners to change the level of care rendered to patients needing ICU level treatment.

There is a subset of advanced cancer patients who benefit greatly from ICU admission and for whom it is appropriate [33], [34]. However, avoiding ICU admission in the last months of life is a key goal for improving patient quality of life [35]. We believe the CARE Track program reduced ICU use through several mechanisms. First, the program established a weekly interdisciplinary meeting (Supportive Care Tumor Board) designed to improve care coordination for CARE Track patients. The meeting provided a routine format in which all disciplines working at the cancer center could discuss patients with worsening symptoms and difficult social situations, as well as identify patients facing imminent transitions of care for discussion. Second, beginning in year 2 of the CARE Track program, patients completed a PRO assessment, with results available from the electronic health record, as a guide to symptom management and decision‐making. Results from the assessment were used to alert clinicians to the needs for symptom control and to anticipate increasing symptom burden for patients in the CARE Track program. Early symptom reporting allows early treatment of developing acute illness, thereby avoiding and reducing the need for subsequent intensive care escalation in many patients. Third, early care planning discussions by palliative care specialists allowed the identification of patients with advanced disease who did not want intensive care or hospitalization at the end of life.

Several limitations attend this research. First, patient allocation was non‐randomized, and selection bias is a potential confounding effect. This observational study relies on multivariable adjustment to account for differences between groups. The CARE Track program was developed as a quality improvement intervention, and implementation within the setting of a randomized trial was not practical. We focused our analysis on concurrent rather than historic controls because the rapidly changing landscape of oncology treatment had the potential to reduce the relevance of comparisons with historic controls. Matching methods were not considered to be reasonable alternatives because of the small study population and the heterogeneous collection of cancer diagnoses.

Several aspects of the analysis reduce the potential effects of selection bias. The multivariable analysis included adjustments for key confounders selected a priori as the most likely influences on the primary outcomes. The large, highly statistically significant adjusted effect sizes demonstrated also reduce the potential effects of selection bias. In each model demonstrating a significant difference between the CARE Track and control groups, the covariate for the intervention effect was the most significant individual covariate.

Another limitation is the potential for multiple comparisons bias. To guard against this, a high threshold p value <.001 was used as the standard for tests of statistical significance for all comparisons.

Conclusion

This study supports the conclusions that an early, concurrent palliative care program significantly reduced ICU admissions at the end of life and led to fewer deaths in the hospital or ICU, but it did not change the in‐ICU experience of patients who require critical care. Patients with advanced cancer who are admitted to the ICU should expect aggressive care, which may be highly beneficial in a subset of patients, but perhaps not desired by those who are less likely to benefit. These data provide further evidence that palliative care exerts the bulk of its influence when applied early in the care of advanced cancer patients and can guide patients’ care to their ultimate goals rather than dealing with the consequences of unintended intensive care admissions.

See http://www.TheOncologist.com for supplemental material available online.

Author Contributions

Conception/Design: Andrew M. Romano, Kristine E. Gade, Robert Havard, James H. Harrison Jr., George J. Stukenborg, Paul W. Read, Leslie J. Blackhall, Patrick M. Dillon,

Provision of study material or patients: James H. Harrison Jr., Paul W. Read, Leslie J. Blackhall, Patrick M. Dillon

Collection and/or assembly of data: Andrew M. Romano, Kristine E. Gade, Gradon Nielsen, Robert Havard, Josh Barclay, Leslie J. Blackhall

Data analysis and interpretation: Andrew M. Romano, Gradon Nielsen, Josh Barclay, Robert Havard, James H. Harrison Jr., George J. Stukenborg, Paul W. Read, Leslie J. Blackhall, Patrick M. Dillon

Manuscript writing: Andrew M. Romano, George J. Stukenborg, Paul W. Read, Leslie J. Blackhall, Patrick M. Dillon

Final approval of manuscript: Andrew M. Romano, James H. Harrison Jr., Paul W. Read, Leslie J. Blackhall, Patrick M. Dillon

Disclosures

Patrick Dillon: Newlink Genetics, Abb‐vie, Novartis, Pfizer (RF). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board