Quality of end-of-life care in low-income, uninsured men dying of prostate cancer

Authors

  • Jonathan Bergman MD,

    Corresponding author
    1. Department of Urology, University of California at Los Angeles, Los Angeles, California
    • Department of Urology, University of California at Los Angeles, Box 951738, Los Angeles, CA 90095-1738
    Search for more papers by this author
    • Fax: (310) 206-5343

  • Amanda C. Chi MS,

    1. David Geffen School of Medicine, University of California at Los Angeles, Los Angeles, California
    Search for more papers by this author
  • Mark S. Litwin MD, MPH

    1. Department of Urology, University of California at Los Angeles, Los Angeles, California
    2. Department of Health Services, University of California at Los Angeles, Los Angeles, California
    3. Jonsson Comprehensive Cancer Center, University of California at Los Angeles, Los Angeles, California
    Search for more papers by this author

Abstract

BACKGROUND:

The quality of end-of-life care was assessed in disadvantaged men prospectively enrolled in a public assistance program. That end-of-life care would be aggressive, moreso than recommended by quality-of-care guidelines, was hypothesized.

METHODS:

Included in the study were all 60 low-income, uninsured men in a state-funded public assistance program who had died since its inception in 2001. To measure quality of end-of-life care, information was collected regarding timing of the institution of new chemotherapeutic regimens, time from administration of last chemotherapy dose to death, the number of inpatient admissions and intensive care unit stays made in the 3 months preceding death, and the number of emergency room visits made in the 12 months before dying. Also noted were hospice use and the timing of hospice referrals.

RESULTS:

Eighteen men (30%) enrolled in hospice before death and the average hospice stay lasted 45 days (standard deviation, 32; range, 2-143 days; median, 41 days). Two patients (11%) were enrolled for fewer than 7 days, and none were enrolled for more than 180 days. The average time from administration of the last dose of chemotherapy to death was 104 days. Chemotherapy was never initiated within 3 months of death, and in only 2 instances (6%) was the final chemotherapeutic regimen administered within 2 weeks of dying. Use of hospital resources (emergency room visits, inpatient admissions, and intensive care unit stays) was uniformly low (mean, 1.0 ± 1.0, 0.65 ± 0.82, and 0.03 ± 0.18, respectively).

CONCLUSIONS:

End-of-life care in disadvantaged men dying of prostate cancer, who enroll in a comprehensive statewide assistance program, is high-quality. Cancer 2010. © 2010 American Cancer Society.

Although a myriad quality-of-care indicators have been developed and validated, their use in analyses of the quality of end-of-life care remain sparse.1-9 Developing methods to care for individuals at the end of life in a high-quality, cost-efficient manner has progressed from a public policy goal to an urgent necessity, because approximately one-third of lifetime Medicare resources are expended during the last year of life.10 For the just‒under 28,000 American men who die of prostate cancer annually, the average cost of care in life's final year triples that of the initial year after diagnosis, in part due to aggressive interventions as death approaches.5, 8, 11 The quality of these individuals' end-of-life care is poorly understood.

Quality of care is the degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge.12 Assessing the quality of end-of-life care previously was considered beyond our ken, due to both the evolving nature of patient expectations at the end of life and the limited measurable outcomes. Recent validation of quality measures now makes that task achievable.1-8 Using these measures, patterns of end-of-life care have been shown to be more aggressive than recommended, especially among disadvantaged individuals.4, 9 Although disadvantaged men typically have worse outcomes than their advantaged counterparts, those enrolled in a comprehensive public assistance program often receive high-quality medical care.13-15 The purpose of the current study was to assess the quality of end-of-life care in low‒income, uninsured men prospectively enrolled in a specific public assistance program.1 Based on previous analyses of administrative data, we hypothesized a priori that end-of-life care would be more inappropriately aggressive than recommended by quality-of-care guidelines.4, 9, 12, 16 Our study was not meant to comprehensively study the quality of end‒of‒life care in all disadvantaged men, but to probe whether high-quality end‒of‒life care could be delivered within a comprehensive cancer care program.

MATERIALS AND METHODS

Subjects

We included all 60 low-income, uninsured men in a state-funded public assistance program who had died since its inception in 2001. IMPACT (Improving Access, Counseling and Treatment for Californians with Prostate Cancer) (available at: www.california-impact.org) was developed to provide access to prostate cancer care for uninsured men with incomes under 200% of the Federal Poverty Level.14, 15 Financially eligible men with biopsy-proven prostate cancer were assigned a primary urology cancer care provider, a clinical care coordinator, and a nurse case manager. They were also connected with counseling services, educational materials, transportation assistance, food security, and housing referrals, and were referred to hospice care when deemed appropriate by their clinical providers. Patients with all tumor grades, disease stages, or history of previous treatment were included. IMPACT covered the costs of all inpatient and outpatient services14, 15 The majority of men served by IMPACT are minorities, younger than most men with prostate cancer, and have prostate disease that is significantly advanced.17

Data Collection

All demographic and clinical information was obtained at enrollment and updated during treatment. The University of California at Los Angeles Office for Protection of Research Subjects approved the data collection protocol, and all procedures were compliant with the Health Insurance Portability and Accountability Act. Data were abstracted from the IMPACT administrative database, which included entries from case managers, clinical care coordinators, and nursing providers. Medical chart abstraction from the IMPACT electronic clinical database and scanned paper records included review of biopsy reports, physicians' documentation of initial and follow-up visits, operative reports, consultation reports, and follow-up encounter forms. The clinical database and nurses' records were cross‒checked to ascertain ethnicity, income, and other demographic data. Data collected included age, ethnicity, job status, partnership status, region of residence (northern vs southern California), and comorbid conditions. We also abstracted tumor grade; clinical disease stage at the time of initial disease diagnosis (not at presentation to the IMPACT program); initial treatment received at IMPACT; subsequent treatment within IMPACT, including zolendroic acid, chemotherapy, and palliative radiotherapy; prostate-specific antigen (PSA) level before initial prostate cancer treatment; date of death; documented PSA closest to time of death; number of emergency room visits made while enrolled in IMPACT; metastasis at time of first presentation to IMPACT; and cause of death. We determined cause of death from the contemporaneous clinical notes from cancer care providers, nurse case managers, and clinical care coordinators.

Quality Indicators

We measured quality of end-of-life care using the Earle quality indicators, which were developed using a modified Delphi approach after focusing on literature review, cancer patient preferences, and expert panel rankings.1 The indicators have been shown to be reliable, valid, and applicable to several patient populations.9 To characterize compliance with the Earle quality indicators, we collected information regarding timing of the institution of new chemotherapeutic regimens, time from the administration of last chemotherapy dose to death, the number of inpatient admissions and intensive care unit stays made in the 3 months preceding death, and the number of emergency room visits made in the 12 months before dying. We also noted patient enrollment in hospice, date of hospice enrollment, and duration of hospice enrollment before death.

RESULTS

Complete demographic and clinical data were available for all 60 subjects at the time of enrollment (Table 1). The majority of men were minorities, most were unemployed, and greater than half were unpartnered. A distinct minority (8%) had D'Amico18 low-risk disease, and nearly three-quarters were treated with androgen deprivation monotherapy within IMPACT; most had evidence of metastases at the time of the initial presentation to IMPACT. Greater than one-third were treated with zolendroic acid, nearly half were palliated with radiotherapy, and greater than half received chemotherapy. The immediate cause of death was prostate cancer in three-quarters of subjects in the current study, whereas the majority of those whose immediate cause of death was not prostate cancer died of cardiac disease. For those whose immediate cause of death was not attributable to prostate cancer, the course of their prostate disease was in its advanced stages.

Table 1. Demographic and Clinical Characteristics of the Study Sample (n = 60)
CharacteristicNo. of Patients (%)
  • IMPACT indicates Improving Access, Counseling and Treatment for Californians with Prostate Cancer; SD, standard deviation; PSA, prostate-specific antigen.

  • a

    Adjuvant treatments were not mutually exclusive, and could therefore sum more than 100%.

Age, y 
 Mean age at IMPACT enrollment (±SD)65.1±11.2
 Mean age at death (±SD)66.4±11.1
Average length of IMPACT enrollment at time of death, mo (±SD)15.0±12.7
Ethnicity 
 White14 (23%)
 African American6 (10%)
 Hispanic39 (65%)
 Asian1 (2%)
Job status 
 Employed23 (38%)
 Unemployed37 (62%)
No. of dependents 
 024 (40%)
 ≥136 (60%)
Partnership status 
 In relationship26 (43%)
 Not in relationship34 (57%)
Region of residence 
 Northern California20 (33%)
 Southern California40 (67%)
Pretreatment PSA level, ng/mL 
 <47 (12%)
 4-106 (10%)
 ≥1047 (78%)
Biopsy Gleason score 
 ≤69 (15%)
 721 (35%)
 ≥830 (50%)
Clinical T classification 
 T1 (localized)23 (38%)
 T2 (localized)21 (35%)
 T3 (locoregional)6 (10%)
 T4 (metastatic)10 (17%)
D'Amico risk stratification 
 Low risk5 (8%)
 Intermediate risk6 (10%)
 High risk49 (82%)
Treatment within IMPACT 
 Radical prostatectomy5 (8%)
 Radiotherapy1 (2%)
 Radiotherapy + androgen deprivation therapy9 (15%)
 Androgen deprivation monotherapy43 (72%)
 Active surveillance2 (3%)
Adjuvant treatmenta 
 Zolendroic acid22 (37%)
 Chemotherapy32 (53%)
 Palliative radiotherapy25 (42%)
PSA level closest to date of death, ng/mL 
 ≤412 (20%)
 4-104 (7%)
 ≥1044 (73%)
Evidence of metastases at presentation to IMPACT43 (72%)
Cause of death 
 Prostate cancer45 (75%)
 Other15 (25%)

Table 2 displays utilization of hospital resources by the subjects in IMPACT. Approximately half made no inpatient stays, and only 2 subjects (3%) were ever admitted to an intensive care unit. The study cohort was divided relatively equally among those making 0, 1, and ≥2 emergency room visits.

Table 2. Utilization of Hospital Resources (n = 60)
Hospital ResourceNo. of Patients
No. of emergency room visits in last 12 mo of life
023 (38%)
120 (33%)
≥217 (28%)
119 (32%)
≥29 (15%)
No. of inpatient stays in last 3 mo of life
032 (53%)
119 (32%)
≥29 (15%)
No. of intensive care unit admissions in last 3 mo of life
058 (97%)
12 (3%)
≥20 (0%)

Compliance with quality indicators was high (Table 3). Eighteen men (30%) enrolled in hospice before death; an additional 5 men (8%) were counseled extensively regarding the hospice option, but declined enrollment. The average hospice stay lasted 45 days (standard deviation, 32 days; range, 2-143 days), with a median of 41 days. Two patients (11%) were enrolled for fewer than 7 days, and none were enrolled for more than 180 days. The average time from administration of the last dose of chemotherapy to death was 104 days. Chemotherapy was never initiated within 3 months of death, and in only 2 instances (6%) was the final chemotherapeutic regimen administered within 2 weeks of dying. Use of hospital resources (emergency room visits, inpatient admissions, and intensive care unit stays) was uniformly low.

Table 3. Compliance with Quality Indicators (n = 60)
Quality IndicatorNo. of Patients
  1. SD indicates standard deviation.

Chemotherapy 
New chemotherapy regimen initiated within 3 mo of death0/32 (0%)
Chemotherapy last administered within 2 wk of death2/32 (6%)
Acute high-intensity end-of-life care 
No. of emergency room visits in last 12 mo of life (mean±SD)1.0 ± 1.0
No. of inpatient admissions in last 3 mo of life (mean±SD)0.65 ± 0.82
No. of intensive care unit stays in last 3 mo of life (mean±SD)0.03 ± 0.18
Hospice 
No. of dying men enrolled in hospice18/60 (30%)
No. of men enrolled for <7 d prior to death2/18 (11%)

DISCUSSION

The current study has 3 important findings. First, patterns of chemotherapy use were appropriate according to the Earle1 quality indicators. No subject initiated a new chemotherapeutic regimen within 3 months of death, and only 2 men (6%) received any chemotherapy within 2 weeks of dying. Only 53% of subjects were treated with chemotherapy overall; although this is consistent with rates of chemotherapy use in other populations, and although it does not represent substandard care according to published indicators, it may suggest potential undertreatment at the beginning of the end-of-life process. Whereas chemotherapy may be used palliatively to improve quality of life for those with incurable disease, administering a new chemotherapeutic regimen near the end of life, or continuing chemotherapy when death is imminent, results in more toxicity than palliation, and is overly aggressive.19 Variation among published indicators exists with regard to delineating the precise timing of chemotherapy that should be used to measure quality of end-of-life care, but the appropriate arc of chemotherapy administration is uniformly accepted.2-8 We used the Earle indicators because they are feasible, valid, and reliable.7 Earle indicators provide a metric with which to measure quality of care at the end of life, and describe what care should and should not be provided as death approaches. However, because quality of end-of-life care is still a maturing field of study, the precise overall compliance rates that are to be expected have yet to be delineated. Retrospective studies of insured individuals, nearly all of whom were white, in both Canada and the United States found that approximately 16% of individuals with cancer continue to receive chemotherapy in the last 2 weeks of life, with a temporal trend toward increased chemotherapy use.4, 9 Black individuals were more likely than whites to receive aggressive chemotherapy.9 That disadvantaged men enrolled in IMPACT used chemotherapy appropriately may reflect the comprehensive counseling they received. Within IMPACT, care is delivered by a multidisciplinary team, including specialists who deliver counseling services and culturally appropriate educational materials. One avenue for improving the quality of end-of-life care may be providing greater guidance regarding disease trajectory to reduce utilization of ineffective, aggressive care as death approaches.

Second, hospital resources, including emergency room visits, inpatient stays, and intensive care unit admissions, were used appropriately by men in our cohort. The median number of emergency room visits made in the last year of life was 1, with just 17 subjects (28%) making ≥2 visits. Few individuals had >1 inpatient stay in the last 3 months of life, and nearly all men were not admitted to an intensive care unit within 3 months of dying. In contrast, studies of the Canadian healthcare system have shown that greater than one-quarter of individuals make at least 1 emergency room visit in the last 2 weeks of life, and 5% are admitted to an intensive care unit within that timeframe.4 Surveillance, Epidemiology, and End Results (SEER)-Medicare analyses of the American population suggest that, in the last month of life, 8% of individuals dying of cancer make ≥2 emergency room visits, 8% make inpatient stays, and an additional 8% are admitted to an intensive care unit facility.9 Such inappropriate use of high-intensity care at the end of life leads to the disproportionate cost of end-of-life care.11

Numerous analyses, including several prospective studies, have shown that nurse visits, home-care support, symptomatic care, and a focus on patient education help prevent the unnecessary use of high-intensity care at the end of life.20-23 IMPACT's comprehensive cancer care model focuses on such value-added services, likely preventing many unnecessary emergency room visits, inpatient stays, and intensive care unit admissions. To reduce the outcome disparities typically observed in low-income and underinsured individuals, IMPACT provides counseling and interpreter services, transportation assistance, and food security and housing referrals, as needed.14, 15 In 2004, The California Dialogue on Cancer released the Comprehensive Cancer Control Plan for California, which calls for “access to quality cancer care” as well as “no disparity in outcomes for all Californians.”24 IMPACT's mission is consistent with California's Comprehensive Cancer Control objective of decreasing cancer-related disparities in low-income, uninsured, and minority men with prostate cancer, and IMPACT delivers the high-quality, end-of-life care demanded by the Comprehensive Cancer Control's objectives. The findings of the current study suggest that the comprehensive model of the IMPACT Program may help deliver high-quality, low-cost care at the end of life and that extending similar comprehensive models to other areas of healthcare need may improve end-of-life care.

Third, the overall rate of hospice utilization was comparable between our cohort and previous descriptions in population-based studies of men dying of prostate cancer, and the timing of the hospice referral compared favorably with previous studies. Approximately one-third of men in the current study enrolled in hospice, only 2 of them (11%) within a week of dying. Previous reports suggest that 18% to 43% of men dying of prostate cancer use hospice, but nearly one-third are enrolled within a week of death.6, 25-28 Both the Institute of Medicine and the American Society of Clinical Oncology stress that hospice programs should aim to palliate disease, not just to manage death.12, 16 In this context, developing methods to reduce inappropriately late hospice referrals are needed; unfortunately, in the past decade, the proportion of individuals referred to hospice very close to death has increased.9 When hospice use is appropriately timed, the cost of end-of-life care is reduced dramatically.29-31 Despite this, and contrary to all published evidence, recent Congressional discussions focused on Medicare reform have considered curtailing the length of hospice stays, presumably to reduce the high cost of end-of-life care.32 In the current study, most men were enrolled in hospice for an appropriate length of time, and the use of costly, high-intensity care at the end of life was low. All men dying of prostate cancer should be counseled about potential hospice use, and individuals who desire hospice care should be enrolled in a timely manner.

The findings of the current study are limited by several methodologic considerations. First, our small samples size may result in our cohort being unrepresentative of other low-income, uninsured men, and our results therefore may not be generalizable. These findings are descriptive and must be validated in a larger population. Despite this limitation, we did demonstrate the feasibility of using the comprehensive cancer-care model to deliver high-quality, end-of-life, care to underserved men. Our findings should not be interpreted to suggest that, as a general rule, low-income, uninsured men receive high-quality care at the end of life. Rather, our study is a proof of principle that, within a comprehensive, IMPACT-type program with many ancillary services, high-quality end-of-life care can be delivered, even to a disadvantaged population. Second, several men in our cohort presented to IMPACT after receiving initial prostate cancer treatment elsewhere, and individuals in our sample had significantly advanced disease at presentation to IMPACT. IMPACT providers therefore had fewer opportunities to address end-of-life goals with their patients than if subjects' disease were not as advanced. Third, the reasons some men who were offered hospice care declined enrollment were unclear. Thorough prospective data will be invaluable in identifying motivations triggering hospice use or nonuse.

Despite these limitations, our methodology strengthens the value of our conclusions. Whereas most end-of-life studies rely on administrative data, we were able to review all nursing and clinical provider notes to glean contemporaneous accounts of patient care. We were also able to identify subjects who declined hospice enrollment despite adequate counseling. Psychosocial care, multidisciplinary treatment, and patient communication have been endorsed as critical components of quality end-of-life care, but to our knowledge are rarely included in quality analyses because of the limitations of administrative data.1 By each of these criteria, IMPACT delivers high-quality care at the end of life. Men enrolled in IMPACT receive a myriad of ancillary services, including psychosocial support and spiritual advisement.14 We did not have specific data regarding pain control and home hospice, both of which would be areas of interest for future study.

End-of-life care in disadvantaged men dying of prostate cancer, who enroll in a comprehensive statewide assistance program, is excellent. The dual influence of comprehensive patient-centered programs on quality and cost must be further validated in larger cohorts, and may serve as a model for efficiently delivering high-quality care at the end of life.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary