CANCER IS A COMMON CAUSE

Transcription

1 ORIGINAL CONTRIBUTION Hospice Use Among Medicare Managed Care and Fee-for-Service Dying With Cancer Ellen P. McCarthy, PhD, MPH Risa B. Burns, MD, MPH Quyen Ngo-Metzger, MD, MPH Roger B. Davis, ScD Russell S. Phillips, MD CANCER IS A COMMON CAUSE of morbidity and mortality among men and women aged 65 years or older. 1 For patients with many types of cancer, effective treatment exists but for others palliative therapies may remain the only option when treatment fails. Decisions to transfer from antitumor therapies to palliative care are complex and are often made only when death is certain. 2,3 The hospice model of care delivers palliative therapies while promoting patient autonomy and decision making at the end of life. 4,5 Although the benefits of hospice are not well-studied, it has become a standard of care for patients with lifethreatening illness. Hospice care has been shown to improve symptom management and quality of life for patients at the end of life. 6,7 Hospice patients are more likely to die at home and are generally more satisfied with their care. 7 Hospice use has increased over time; however, many patients enroll too late to maximize the benefits of hospice services The Medicare hospice benefit, established by Congress in 1982, provides families with resources to care for their dying loved one at home. 14 The Medicare hospice benefit is one of only a few Context For most patients aged 65 years or older with cancer, hospice services are uniformly covered by Medicare. Hospice care is believed to improve care for patients at the end of life. However, few patients use hospice and others enroll too late to maximize the benefits of hospice services. Objectives Because type of insurance may affect use, we examined whether patients with Medicare managed care insurance enrolled in hospice earlier and had longer hospice stays than patients with Medicare fee-for-service (FFS) insurance. Design and Setting Retrospective analysis of the last year of life using the Linked Medicare-Tumor Registry Database in 1 of 9 Surveillance, Epidemiology, and End Results program coverage areas. A total of Medicare beneficiaries aged 66 years or older diagnosed with first primary lung (n=62117), colorectal (n=57260), prostate (n=59826), female breast (n=37609), bladder (n=19598), pancreatic (n=11378), gastric (n=9599), or liver (n=2703) cancer between January 1, 1973, and December 31, 1996, and who died between January 1, 1988, and December 31, Main Outcome Measures Time from diagnosis to hospice entry and hospice length of stay for patients enrolled in FFS vs managed care plans after adjusting for patient demographics, tumor registry, year of hospice entry, and type and cancer stage. Results Of the patients, most were men (59%), white (85%), and enrolled in FFS (89.7%). Only patients (21.1%) received hospice care before death. Hospice use varied by type of primary cancer ranging from 31.8% of patients with pancreatic cancer to 15.6% with bladder cancer. Managed care patients were more likely to use hospice than FFS patients (32.4% vs 19.8%, P.001). Among hospice patients, median (interquartile range) length of stay was longer for managed care vs FFS patients (32 days [11-82] vs 25 days [9-66], P.001). After adjustment, managed care patients had higher rates of hospice enrollment (adjusted hazard ratio [HR], 1.38; 95% CI, ) and had a longer length of stay (adjusted HR, 0.91; 95% CI, ) vs FFS patients. Managed care patients were less likely to enroll in hospice within 7 days of their death (18.6% vs 22.6%, P.001) and somewhat more likely to enroll in hospice more than 180 days before death (7.8% vs 6.1%, P.001); the results for each of the 8 cancer diagnoses were similar. Hospice enrollment and length of stay among managed care vs FFS patients differed significantly by region. Conclusion Medicare beneficiaries enrolled in managed care had consistently higher rates of hospice use and significantly longer hospice stays than those enrolled in FFS. Although these differences may reflect patient and family preferences, our findings raise the possibility that some managed care plans are more successful at facilitating or encouraging hospice use for patients dying with cancer. JAMA. 2003;289: Author Affiliations: Department of Medicine, Harvard Medical School, Beth Israel Deaconess Medical Center, Boston, Mass. Corresponding Author and Reprints: Ellen P. McCarthy, PhD, MPH, Department of Medicine, Harvard Medical School, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Rose-139, Boston, MA ( ellen_mccarthy@bidmc.harvard.edu) JAMA, May 7, 2003 Vol 289, No. 17 (Reprinted) 2003 American Medical Association. All rights reserved.

2 services carved out of managed care. This benefit is unique because the Centers for Medicare and Medicaid Services reimburse hospices directly for services provided to managed care beneficiaries. 15 Once a beneficiary enrolls in hospice, the managed care organization is relieved of its responsibility for potentially expensive end of life care. 15 Higher rates of hospice use among Medicare beneficiaries with managed care insurance were first described by Virnig et al 12 for beneficiaries residing in South Florida in 1992 and in 94 of 100 counties examined across the United States during In this context, we used the Linked Medicare-Tumor Registry Database 17 to examine the relationship between Medicare managed care and hospice use among patients dying with cancer from 1988 to Specifically, we sought to determine whether Medicare beneficiaries with managed care insurance had higher rates of hospice enrollment and longer hospice stays than beneficiaries with fee-for-service (FFS) insurance after adjusting for patient demographics, cancer characteristics, and illness duration. We further examined whether these relationships were consistent across 8 cancer diagnoses with varying prognoses and across 9 geographically diverse regions with varying levels of managed care penetration. METHODS Data Source We conducted a retrospective analysis of the last year of life using the Linked Medicare-Tumor Registry Database. 17 The linked database contains cancer information on patients aged 65 years or older from the National Cancer Institute s Surveillance, Epidemiology, and End Results (SEER) program and Medicare enrollment and use information from the Centers for Medicare and Medicaid Services. Medicare files are available for the years 1986 to 1998 for patients diagnosed with cancer between 1973 and Detailed information describing the linkage and match rates between SEER and Medicare is published elsewhere. 17 We used the Medicare denominator file and hospice file. The denominator file is generated annually and contains specific demographic and enrollment information, including date of death, for every Medicare beneficiary. The hospice file contains claims for every beneficiary who received hospice services under the Medicare hospice benefit, including services provided to beneficiaries enrolled in managed care. Study Sample We studied Medicare beneficiaries who were diagnosed with cancer between January 1, 1973, and December 31, 1996, who resided in 1 of the 9 SEER coverage areas, and who died between January 1, 1988, and December 31, To ensure that beneficiaries had Medicare coverage during their last year of life, we limited our sample to patients diagnosed with cancer at 66 years or older. We studied men and women with a first primary diagnosis of lung (n=62117), colorectal (n=57260), prostate (n=59826), female breast (n=37609), bladder (n=19598), pancreatic (n=11378), gastric (n=9599), or liver (n=2703) cancer. These diagnoses include the 5 most commonly diagnosed cancers in elderly individuals and cancers disproportionately diagnosed in certain racial or ethnic minority groups. Outcomes of Interest We examined 2 primary outcome measures. Hospice enrollment was measured from patients date of diagnosis to hospice entry or death. This analysis used the cancer patients in our study to compare the rate of hospice enrollment between managed care and FFS patients. who died without receiving hospice care were treated as censored observations (n=205153). Length of stay in hospice was measured from date of hospice enrollment until discharge or death, whichever came first. This analysis focused on patients who received hospice care during the last year of life. We excluded 400 patients whose date of death predated their last hospice service, because we were unable to reconcile this discrepancy. Therefore, the final sample for the length of stay analysis consisted of hospice patients. who left hospice before they died were treated as censored observations (n=6743). Managed Care Enrollment We used enrollment information from the denominator file to identify beneficiaries enrolled in managed care organizations. Medicare beneficiaries can change their insurance plan on a monthly basis. Therefore, the denominator file contains an indicator of managed care enrollment for each month of the year. We classified beneficiaries as having FFS insurance if they were continuously enrolled in FFS throughout the last 6 months of life (n=233238). We classified beneficiaries as having managed care insurance if they were enrolled in managed care at anytime during the last 6 months of life (n=26852); however, 92% of beneficiaries with managed care insurance were continuously enrolled in managed care throughout the last 6 months of life. Covariates We obtained the following information: sex, race (non-hispanic white, black, other), marital status at diagnosis (married, not married), geographic area of residence (urban, rural), and SEER tumor registry. We categorized age at diagnosis (range, years) as 66 to 69, 70 to 74, 75 to 79, 80 to 84, and 85 years or older. Because individual patient incomes were not available, we used ZIP code level census data to group patients into quintiles of median household income. We classified stage using the historical system (local, regional, distant, and unstaged) for 2 reasons. First, SEER has collected the historical staging system since its inception in 1973; therefore, it is available for all patients, and SEER began collecting the American Joint Committee on Cancer (AJCC) staging system in 1988; however, it is only available for very few patients diag American Medical Association. All rights reserved. (Reprinted) JAMA, May 7, 2003 Vol 289, No

3 nosed before Second, the AJCC system is not available for patients with pancreatic, gastric, or liver cancer (3 of the 8 cancer diagnoses studied). We repeated our analyses using the AJCC system for patients with lung, colorectal, prostate, breast, and bladder cancer and found similar results. Statistical Analyses All statistical analyses were performed using SAS software version 6.12 (SAS Institute, Cary, NC). P.05 was considered statistically significant. We performed bivariable analyses to determine hospice use across sociodemographic characteristics, type of health insurance, stage at diagnosis, and year of diagnosis. 2 Statistics and t tests were used to identify factors that were significantly associated with hospice use. For hospice patients, we used Kaplan-Meier analysis to estimate median length of stay comparing managed care and FFS patients and used log-rank tests to identify significant differences in hospice length of stay. We examined the distribution of length of stay by type of health insurance. Specifically, we examined 3 potential indicators of quality of care. First, we compared proportions of patients with managed care and FFS on hospice enrollment within 7 days of death as it may be an indicator of poor quality of care (ie, patients enrolled too late for maximum benefit). 11 Second, we examined hospice enrollment more than 180 days before death because it may indicate inappropriate use of hospice services. 11 Third, we sought to examine an indicator of the appropriate amount of time in hospice required to derive maximal benefits. Because there is no set standard of care, we examined hospice enrollment for 2 months or longer because it had been used previously 18 (results were similar when greater enrollment for 3 months or longer was used). We fit multivariable Cox proportional hazards regression models for each outcome to determine whether differences in hospice use between patients with managed care insurance and FFS insurance persist after adjusting for demographic and clinical characteristics. We adjusted for the following demographic (sex, race, and marital status at diagnosis, residence in urban or rural area, and median household income of ZIP code of residence) and clinical (stage at diagnosis and type of primary cancer) factors. Additionally, we adjusted for the effects of the different locations of the SEER tumor registries using a Cox proportional hazards regression model that allowed a different underlying hazard for each tumor registry. The hospice enrollment model also adjusted for age and year of diagnosis. The hospice length of stay model also adjusted for age at hospice entry, year of hospice entry, and illness duration (measured as time from diagnosis until hospice entry). Models were fit for all cancer diagnoses combined and for each type of primary cancer separately. To explore whether the effect of managed care persisted among patients who upon diagnosis were clinically appropriate for hospice, we repeated our analyses on patients who were diagnosed with stage IV lung and colorectal cancer or with distant pancreatic, liver, and gastric cancer. Because roughly 75% of patients enrolled in managed care came from tumor registries located in northern California, Seattle, and Hawaii, we further explored the effect of managed care penetration and geographic location on hospice use by examining the relationship between hospice enrollment and length of stay within each of the 9 SEER registries included in our study. We estimated adjusted hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) from the coefficient and SE of the Cox proportional hazards regression model. 19 Adjusted HRs greater than 1.0 signify higher rates of hospice enrollment and adjusted HRs of less than 1.0 indicate longer hospice stays among patients with Medicare managed care insurance due to a lower rate of discharge, which translates to a shorter length of stay. RESULTS TABLE 1 presents demographic and cancer characteristics of the patients studied. Overall, patients (10.3%) were enrolled in Medicare managed care. TABLE 2 presents differences in hospice use between patients with managed care and FFS insurance. Overall, patients (21.1%) used hospice. with managed care insurance were more likely to use hospice services than patients with FFS insurance (8698 [32.4%] vs [19.8%], respectively). Although hospice use varied widely across the different cancer diagnoses, patients with managed care insurance were consistently more likely to receive hospice care than patients with FFS insurance. The FIGURE illustrates Medicare managed care enrollment and hospice use during 11 years for our study sample. As expected, the proportion of patients enrolled in Medicare managed care increased during the study years. From 1988 to 1998, hospice use increased for both managed care and FFS patients; however, during any given year hospice use was greater among managed care patients. TABLE 3 presents length of stay in hospice for patients with managed care and FFS insurance as well as the distribution of time in hospice. There was a 1-week difference in median (interquartile range) length of stay between managed care and FFS patients (32 days [11-82] vs 25 days [9-66], respectively, P.001). Managed care patients were less likely to enroll in hospice within 7 days of death compared with FFS patients (1595 [18.6%] vs [22.6%], respectively; P.001). This finding was consistent across cancer diagnoses. We also found that overall managed care patients were somewhat more likely to enroll in hospice more than 180 days before death (668 [7.8%] vs 2816 [6.1%]; P.001). However, within specific cancer diagnoses, only those managed care patients with lung, colorectal, or prostate cancer were significantly more likely to enroll in hospice more than 180 days before death. Finally, only 25% of patients dy JAMA, May 7, 2003 Vol 289, No. 17 (Reprinted) 2003 American Medical Association. All rights reserved.

4 Table 1. Patient Demographic and Cancer Characteristics of the Study Sample by Type of Primary Cancer No. (%) Time From Diagnosis Until Death Type of Enrolled in Age at Race Stage at Diagnosis Died 6 Died 12 Primary Cancer No. of Managed Care, No. (%) Diagnosis, Mean (SD), y Men White Black Local Distant Months After Diagnosis Months After Diagnosis All patients (10.3) 76 (7) (59) (85) (8) (39) (24) (22) (68) Lung (10.4) 74 (6) (61) (85) 4976 (8) 9443 (15) (46) (44) (35) Colorectal (9.9) 77 (7) (48) (86) 3802 (7) (35) (20) 9860 (17) (74) Prostate (10.7) 76 (6) (100) (83) 6010 (10) (58) (17) 3074 (5) (89) Breast (9.7) 76 (7) (90) 2276 (6) (54) 3205 (9) 1951 (5) (90) Bladder (10.2) 77 (7) (73) (92) 788 (4) (69) 751 (4) 2027 (10) (81) Pancreatic (10.9) 76 (7) 5055 (44) 9250 (81) 1072 (9) 976 (9) 5147 (45) 7390 (65) 1809 (16) Gastric (12.6) 77 (7) 5772 (60) 6825 (71) 1009 (11) 1937 (20) 3141 (33) 3998 (42) 3952 (41) Liver (12.9) 75 (6) 1769 (66) 1822 (67) 260 (10) 651 (24) 567 (21) 1815 (67) 462 (17) ing with cancer used hospice for 2 months or more. However, overall, managed care patients were more likely than FFS patients to use hospice for at least 2 months before death (2460 [28.6%] vs [23.1%], respectively; P.001). This finding was consistent for patients dying with lung, colorectal, prostate, pancreatic, and gastric cancers. TABLE 4 presents results from multivariable analyses of hospice enrollment and length of stay in hospice for all patients and for each type of cancer. Overall, patients with managed care insurance had higher rates of hospice use than patients with FFS insurance (adjusted HR, 1.38; 95% CI, ). Rates of hospice enrollment were consistently higher among managed care patients for each type of cancer. Among hospice patients, those with managed care insurance had longer stays in hospice than patients with FFS insurance (adjusted HR, 0.91; 95% CI, ). We found consistent results among patients with lung, colorectal, and prostate cancer. Moreover, we found even stronger associations between managed care and hospice enrollment (adjusted HR, 1.49; 95% CI, ) and length of stay (adjusted HR, 0.88; 95% CI, ) among cancer patients who, at the time of diagnosis, had stage IV or metastatic disease. TABLE 5 presents results of multivariable analyses of hospice enroll- Table 2. Bivariable Comparison of Hospice Use Between Fee-for-Service and Managed Care by Type of Primary Cancer* Type of Primary Cancer Figure. Medicare Managed Care Enrollment and Hospice Use Among Dying With Cancer % No. of Enrolled in Hospice, No. (%) All Fee-for-Service Managed Care All patients (21.1) (19.8) 8698 (32.4) Lung (27.0) (25.4) 2596 (40.1) Colorectal (19.8) 9605 (18.6) 1727 (30.4) Prostate (18.1) 8992 (16.8) 1858 (29.0) Breast (16.4) 5293 (15.5) 880 (25.2) Bladder (15.6) 2585 (14.7) 477 (23.9) Pancreatic (31.8) 3309 (30.0) 576 (46.3) Gastric (24.7) 1946 (23.2) 427 (35.3) Liver (28.9) 625 (26.5) 157 (45.1) *All bivariable comparisons between fee-for-service and managed care patients were significant at P.001. Hospice Use, % Managed Care (n = ) Fee-for-Service (n = ) Enrolled in Managed Care, % (N = ) Year of Death Proportion of patients enrolled in managed care and in hospice were computed by year of death American Medical Association. All rights reserved. (Reprinted) JAMA, May 7, 2003 Vol 289, No

5 ment and length of stay in hospice for each tumor registry. Hawaii and San Francisco/Oakland had the greatest proportion of patients enrolled in Medicare managed care. San Francisco/ Oakland and Seattle/Puget Sound had the strongest associations between enrollment in managed care and rates of Table 3. Bivariable Comparison of Length of Stay in Hospice Between Fee-for-Service and Managed Care by Type of Primary Cancer* Type of Primary Cancer No. of Fee-for-Service Managed Care P Value Length of Stay in Hospice, Median (Interquartile Range) All hospice patients (9-66) 32 (11-82).001 Lung (9-62) 34 (11-87).001 Colorectal (10-69) 35 (12-87).001 Prostate (9-74) 31 (10-91).001 Breast (8-76) 30 (10-77).61 Bladder (6-50) 21 (8-53).49 Pancreatic (9-62) 30 (14-83).003 Gastric (9-55) 28 (11-66).08 Liver (7-65) 22 (8-55).77 Hospice Enrolled Within 7 Days of Death, No. (%) All hospice patients (22.6) 1595 (18.6).001 Lung (22.2) 468 (18.2).001 Colorectal (20.9) 281 (16.5).001 Prostate (22.8) 376 (20.5).03 Breast (23.4) 168 (19.4).01 Bladder (29.8) 111 (23.5).006 Pancreatic (21.5) 86 (15.1).001 Gastric (21.2) 72 (16.9).04 Liver (25.9) 33 (21.4).26 Hospice Enrolled 180 Days Before Death, No. (%) All hospice patients (6.1) 668 (7.8).001 Lung (5.4) 201 (7.8).001 Colorectal (6.5) 149 (8.7).001 Prostate (7.0) 164 (8.9).003 Breast (8.4) 74 (8.5).90 Bladder (4.3) 16 (3.4).37 Pancreatic (4.7) 33 (5.8).27 Gastric (4.4) 27 (5.6).27 Liver (6.0).49 Hospice Enrolled for 2 Months, No. (%) All hospice patients (23.1) 2460 (28.6).001 Lung (21.6) 777 (30.3).001 Colorectal (24.4) 517 (30.3).001 Prostate (25.4) 552 (30.1).001 Breast (26.0) 240 (27.7).29 Bladder (17.5) 83 (17.6).98 Pancreatic (21.3) 161 (28.1).001 Gastric (19.1) 102 (23.9).02 Liver (21.5) 28 (18.2).37 *Percentages were computed by number of hospice patients enrolled divided by total number of patients in hospice by type of cancer and type of insurance. P values are bivariable comparison between fee-for-service and managed care patients. Less than 10 patients. hospice enrollment. Hawaii, San Francisco/Oakland, and New Mexico were the only regions in which managed care enrollment was significantly associated with hospice length of stay. As the proportion of beneficiaries enrolled in managed care decreased, rates of hospice use and length of stay were similar except that patients with managed care insurance had lower rates of hospice enrollment in metropolitan Atlanta than FFS patients and shorter hospice stays in Connecticut. COMMENT This is the first population-based study to our knowledge to examine the relationship between Medicare managed care and hospice entry and length of stay in hospice among patients with 8 different cancer diagnoses in 9 geographically diverse regions across the United States during an 11-year period. We found that although hospice use varied across different cancer diagnoses, patients enrolled in Medicare managed care had consistently higher rates of hospice use than patients enrolled in Medicare FFS. We also found overall that patients with managed care insurance had significantly longer hospice stays than patients with FFS insurance. Although the median length of stay was higher among patients with managed care insurance for each of the 8 primary cancer types, statistical significance was not achieved for some diagnoses. We found that patients with managed care insurance were less likely to enroll in hospice care within 7 days of death and more likely to be enrolled in hospice for 2 or more months before death than patients with FFS insurance. These findings suggest that patients with managed care insurance have a greater opportunity to benefit from hospice care. Admission to hospice within a week of death does not give hospice providers sufficient time to fully develop and implement a patient-centered care plan or to adequately prepare families to care for their loved one at home. 10,11 The optimal amount of time in hospice required for patients to derive maximum benefit is not well-studied. Some experts suggest that patients should use hospice for at least 2 months 18 preceding death while others suggest at least 3 months. 20 Our study shows that even though patients with managed care insurance have a distinct advantage over patients with FFS insurance in enroll JAMA, May 7, 2003 Vol 289, No. 17 (Reprinted) 2003 American Medical Association. All rights reserved.

6 ing in hospice and staying longer, few patients received hospice care for 2 months or more. In fact, the majority of hospice patients received hospice care for 1 month or less whether they were enrolled in managed care or FFS. Different financial incentive structures between managed care and FFS systems are one explanation for higher rates of hospice use among Medicare managed care beneficiaries. 12,16,21,22 Because the Medicare program assumes the costs of care associated with the lifethreatening illness for beneficiaries who enroll in hospice, a potential financial incentive exists for managed care organizations to transfer dying patients to hospice care to avoid paying the high costs of care at the end of life. We explored the possibility that patients with managed care insurance were enrolled in hospice inappropriately. We found that although patients with managed care insurance were somewhat more likely to enroll in hospice more than 180 days before death than patients with FFS, the absolute difference was negligible. Additionally, we found that the positive association with managed care enrollment was much more pronounced among patients with cancer who were clinically appropriate for hospice entry at the time of diagnosis, such as patients diagnosed with stage IV cancer or distant metastases. Although it is possible that financial incentives are aligned to encourage hospice use for managed care patients, 23 we found no evidence to suggest that managed care organizations are conserving their resources by enrolling patients in hospice care inappropriately. Finally, we found that the association of managed care with hospice use was not consistent across regions of the United States but seemed to be related to managed care enrollment in Hawaii and San Francisco/ Oakland. These data suggest that factors other than financial incentives may be responsible for the associations. Our findings are consistent with those of Virnig et al, 12,16 who found that decedents who were enrolled in managed care had higher rates of hospice use and longer hospice stays than those enrolled in FFS in South Florida and in several counties across the United Sates. Our results complement those of Virnig et al 12,16 by examining time from diagnosis to hospice for patients with Table 4. Adjusted Hazard Ratios for Hospice Enrollment and Length of Stay in Hospice Type of Primary Cancer Sample Used in Analysis Hospice Enrollment* Adjusted Hazard Ratio for Managed Care (95% CI) Length of Stay in Hospice Sample Used in Analysis Adjusted Hazard Ratio for Managed Care (95% CI) All patients ( ) ( ) Lung ( ) ( ) Colorectal ( ) ( ) Prostate ( ) ( ) Breast ( ) ( ) Bladder ( ) ( ) Pancreatic ( ) ( ) Gastric ( ) ( ) Liver ( ) ( ) Abbreviation: CI, confidence interval. *Hospice enrollment: adjusted hazard ratio 1.00 indicates higher rates of hospice enrollment for managed care patients. All models are adjusted for age, sex, race, income quintile, marital status, residence in an urban or rural area, stage at diagnosis, and Surveillance, Epidemiology, and End Results (SEER) tumor registry. These analyses were performed on all patients. Length of stay in hospice: adjusted hazard ratio 1.00 indicates longer stays in hospice for managed care patients. All models are adjusted for age, sex, race, income quintile, marital status, residence in an urban or rural area, stage at diagnosis, time from diagnosis to hospice enrollment, and SEER tumor registry. These analyses were performed only on patients who enrolled in hospice care. Reference group for adjusted hazard ratio is fee-for-service patients. Models for all patients are also adjusted for type of cancer. Table 5. Comparison of Managed Care and Fee-for-Service by SEER Tumor Registry for All Cancer Diagnoses Combined* SEER Tumor Registry No. of Enrolled in Managed Care, No. (%) Enrolled in Hospice, No. (%) Hospice Enrollment, Adjusted HR (95% CI) Length of Stay in Hospice, Adjusted HR (95% CI) Hawaii (35) 2207 (22) 1.15 ( ) 0.86 ( ) San Francisco/Oakland (29) 8529 (21) 1.53 ( ) 0.84 ( ) Seattle/Puget Sound (14) 7761 (21) 1.58 ( ) 0.96 ( ) New Mexico (13) 3303 (26) 1.00 ( ) 0.86 ( ) Utah (6) 1312 (12) 0.87 ( ) 0.93 ( ) Iowa (4) 8263 (20) 1.16 ( ) 1.06 ( ) Connecticut (3) 7883 (18) 1.04 ( ) 1.15 ( ) Metropolitan Atlanta (2) 3167 (21) 0.62 ( ) 0.83 ( ) Metropolitan Detroit (2) (26) 1.03 ( ) 0.95 ( ) Abbreviations: CI, confidence interval; HR, hazard ratio; SEER, Surveillance, Epidemiology, and End Results. *Reference group for adjusted HRs is fee-for-service patients. Hospice enrollment: adjusted HR 1.00 indicates higher rates of hospice enrollment for managed care patients. All models are adjusted for age, sex, race, income quintile, marital status, residence in an urban or rural area, stage at diagnosis, and type of cancer. These analyses were performed on all patients within a registry (sample sizes located in column 2). Length of stay in hospice: adjusted HR 1.00 indicates longer stays in hospice for managed care patients. All models are adjusted for age, sex, race, income quintile, marital status, residence in an urban or rural area, stage at diagnosis, time from diagnosis to hospice enrollment, and type of cancer. These analyses were performed only on patients who enrolled in hospice care within a registry (sample sizes located in column 4) American Medical Association. All rights reserved. (Reprinted) JAMA, May 7, 2003 Vol 289, No

7 different cancer diagnoses residing in several US regions. Moreover, we found that the relationship between managed care and hospice use persisted after adjusting for patient demographics, cancer stage at diagnosis, duration of cancer illness, and the location of the SEER tumor registry. Very little is known about the differences in care experienced by patients with managed care and FFS insurance at the end of life. Cher and Lenert 24 found that patients with Medicare managed care were less likely than patients with FFS to receive potentially ineffective care (ie, aggressive critical care treatments that are expensive but have an unclear benefit). It is possible that elderly patients who would prefer a more aggressive course of end of life care would be more apt to participate in the FFS option. 25 Studies show that managed care discourages the participation of seriously ill persons, especially those with multiple chronic conditions The Medical Outcomes Study 30 demonstrated that chronically ill elderly patients enrolled in managed care experienced poorer health outcomes than patients in FFS after 4 years of follow-up. Although previous research shows that Medicare managed care beneficiaries are more likely to disenroll from managed care when they are sick, 29 the vast majority of patients in our study remained in managed care throughout the last 6 months of life. The 8% of patients who disenrolled from managed care at the end of life had demographic characteristics similar to those who remained in managed care. They were, however, less likely to enroll in hospice (data not presented); therefore, our associations are underestimated and conservative. Although market penetration of managed care appears to have a substantial effect on hospice use, the generalizability of this finding is questionable. Associations between managed care and hospice enrollment and length of stay were strong in West coast regions populated by a large proportion of beneficiaries enrolled in managed care. These regions are predominantly served by large staffmodel managed care organizations, such as Kaiser Permanente and Group Health Cooperative. However, Emanuel et al 23 found higher hospice rates among Medicare managed care beneficiaries in Massachusetts, a state with a lower managed care penetration and a higher mix of organizational models (eg, group or staff, independent practice associations) of managed care than California. We found that hospice use was similar in most areas with few managed care enrollees suggesting perhaps that some level of managed care penetration is necessary for hospice enrollment practices to differ between managed care and FFS systems. Moreover, most studies of managed care examine managed care as a single entity. Our data do not allow us to identify the managed care organizations in which patients were enrolled. Therefore, we were unable to examine how differences in organizational models and profit status relate to increased hospice use. We were also unable to explore relationships between managed care and hospice organizations. For example, some managed care organizations have established formal affiliations with organizations that provide hospice care. 31 Others, such as Kaiser Permanente, own or operate their own hospices. It is important to note other potential limitations to our study. Most importantly, we did not have information about patients preferences for care. Medicare beneficiaries who enroll in managed care may be systematically different from those who select FFS. 32 It is possible that dying patients enrolled in managed care might favor a less aggressive care plan. Second, we lacked factors that could potentially influence patients care at the end of life. Specifically, we were unable to adjust for patient comorbidity, because information on comorbid illness was not available for patients with managed care insurance. Third, we lack factors related to the health care system including local availability of hospices. Finally, we had no information about patients physicians, including their specialty, their experience caring for dying patients, or their practice patterns. Our study demonstrates that, in general, Medicare beneficiaries with managed care insurance access hospice care earlier and stay longer than patients with FFS insurance. Moreover, areas with high market penetration of managed care appear to be responsible for this association. We found that managed care patients, particularly those residing in market areas dominated by managed care, are significantly more likely than FFS patients to access hospice care at the end of life. Our study raises the possibility that managed care organizations are more successful at facilitating or encouraging hospice use for patients dying with cancer. Additional research is needed to determine the mechanisms by which this occurs. Further research should evaluate how the observed differences in hospice use relate to patients preferences for care, physician practice patterns, financial incentives, and system differences in the delivery of care at the end of life. Although our findings may not represent a general effect of managed care and in fact may represent the practices of a few managed care organizations with effectively integrated health care systems or established relationships with hospice providers, it would be worthwhile to explore whether elements of managed care responsible for increased hospice use can be integrated into the FFS system to facilitate access to hospice care for dying patients. However, the opportunity to learn from managed care providers is narrowing, because the availability of Medicare managed care has declined substantially since Author Contributions: Study concept and design: McCarthy, Burns, Davis, Phillips. Acquisition of data: McCarthy. Analysis and interpretation of data: McCarthy, Burns, Ngo-Metzger, Davis, Phillips. Drafting of the manuscript: McCarthy. Critical revision of the manuscript for important intellectual content: McCarthy, Burns, Ngo-Metzger, Davis, Phillips. Statistical expertise: McCarthy, Davis. Obtained funding: McCarthy. Administrative, technical, or material support: McCarthy, Burns, Ngo-Metzger. Study supervision: Burns, Phillips. Funding/Support: Dr McCarthy is the recipient of First Independent Research Support and Transition (FIRST) Award 5R29 CA79052, which is funded by the National Cancer Institute. Disclaimer: This study used the Linked Medicare JAMA, May 7, 2003 Vol 289, No. 17 (Reprinted) 2003 American Medical Association. All rights reserved.

8 Tumor Registry Database. The interpretation and reporting of these data are the sole responsibility of the authors. Previous Presentation: Presented in part at the National Meeting of the Society of General Internal Medicine, May 3, 2002, Atlanta, Ga. Acknowledgment: We thank several groups for their efforts in creating and disseminating the Linked Medicare-Tumor Registry Database, including the Applied Research Branch, Division of Cancer Prevention and Population Science, National Cancer Institute; the Office of Information Services and the Office of Strategic Planning, Centers for Medicare and Medicaid Services (formerly the Health Care Financing Administration); Information Management Services Inc; and the Surveillance, Epidemiology, and End Results program tumor registries. REFERENCES 1. American Cancer Society: Cancer Facts and Figures Atlanta, Ga: American Cancer Society; Cleary JF, Carbone PP. Palliative medicine in the elderly. Cancer. 1997;80: Levy MH. Living with cancer: hospice/palliative care. J Natl Cancer Inst. 1993;85: Bulkin W, Lukashok H. Rx for dying: the case of hospice. N Engl J Med. 1988;318: Kinzbrunnner BM. Hospice: 15 years and beyond in the care of the dying. J Palliat Med. 1998;1: Wallston KA, Burger C, Smith RA, Baugher RJ. Comparing the quality of death for hospice and nonhospice cancer patients. Med Care. 1988;26: Greer DS, Mor V, Morris JN, Sherwood S, Kidder D, Birnbaum H. An alternative in terminal care: results of the National Hospice Study. J Chronic Dis. 1986;39: General Accounting Office. Medicare: More Beneficiaries Use Hospice But for Fewer Days. Washington, DC: General Accounting Office; Publication GAO/HEHS Available at: Accessed March 6, National Hospice and Palliative Care Organization. Delivering quality care and cost-effectiveness at the end of life: building on the 20-year success of the Medicare Hospice Benefit, February Available at: Accessed March 6, Christakis NA. Timing of referral of terminally ill patients to an outpatient hospice. J Gen Intern Med. 1994;9: Christakis NA, Escarce JJ. Survival of Medicare patients after enrollment in hospice programs. N Engl J Med. 1996;335: Virnig BA, Persily NA, Morgan RO, DeVito CA. Do Medicare HMOs and Medicare FFS differ in their use of the Medicare Hospice Benefit? Hosp J. 1999;14: McCarthy EP, Burns RB, Davis RB, Phillips RS. Barriers to hospice care among older patients dying with lung and colorectal cancer. J Clin Oncol. 2003;21: Centers for Medicare and Medicaid Services. The Medicare Hospice Benefit. Baltimore, Md: US Dept of Health and Human Services; Publication HCFA Available at: /publications/pubs/pdf/02154.pdf. Accessed March 6, Centers for Medicare and Medicaid Services. Hospice Manual: Chapter 2: Coverage of Services. Baltimore, Md: US Dept of Health and Human Services; Publication 21. Available at: Accessed March 6, Virnig BA, Fisher ES, McBean AM, et al. Hospice use in Medicare managed care and fee-for-service systems. Am J Manag Care. 2001;7: Potosky AL, Riley GF, Lubitz JD, et al. Potential for cancer related health services research using a Linked Medicare-Tumor Registry Database. Med Care. 1993;31: Last Acts. Means to a better end: a report on dying in America today. Available at: Accessed March 6, Lee ET. Statistical Methods for Survival Data Analysis. 2nd ed. New York, NY: John Wiley & Sons Inc; Lamont EB, Christakis NA. Physician factors in the timing of cancer patient referral to hospice palliative care. Cancer. 2002;94: Sulmasy DP. Managed care and managed death. Arch Intern Med. 1995;155: Miles SH, Weber EP, Koepp R. End-of-life treatment in managed care: the potential and the peril. West J Med. 1995;163: Emanuel EJ, Ash A, Yu W, et al. Managed care, hospice use, site of death, and medical expenditures in the last year of life. Arch Intern Med. 2002;162: Cher DJ, Lenert LA. Method of Medicare reimbursement and the rate of potentially ineffective care of critically ill patients. JAMA. 1997;278: Curtis JR, Rubenfeld GD. Aggressive medical care at the end of life: does capitated reimbursement encourage the right care for the wrong reason? JAMA. 1997;278: Clement DG, Retchin SM, Brown RS, Stegall MH. Access and outcomes of elderly patients enrolled in managed care. JAMA. 1994;271: Steinberg SM, Block SD. Caring for patients at the end of life in an HMO. J Palliat Med. 1998;1: Neuman P, Maibach E, Dusenbury K, Kitchman M, Zupp P. Marketing HMOs to Medicare beneficiaries. Health Aff (Millwood). 1998;17: Morgan RO, Virnig BA, DeVito CA, Persily NA. The Medicare-HMO revolving door: the healthy go in and the sick go out. N Engl J Med. 1997;337: Ware JE Jr, Bayliss MS, Rogers WH, Kosinski M, Tarlov AR. Differences in 4-year health outcomes for elderly and poor, chronically ill patients treated in HMO and fee-for-service systems: results from the Medical Outcomes Study. JAMA. 1996;276: Randal J. Hospice services feel the pinch of managed care. J Natl Cancer Inst. 1996;88: Riley G, Tudor C, Chiang YP, Ingber M. Health status of Medicare enrollees in HMOs and fee-forservice in Health Care Financ Rev. 1996;17: Centers for Medicare and Medicaid Services. Fact Sheet: Protecting Medicare Beneficiaries When Their Medicare + Choice Organization Withdraws. Baltimore, Md: Centers for Medicare and Medicaid Services; Available at: /media/press/release.asp?counter=638. Accessed March 6, American Medical Association. All rights reserved. (Reprinted) JAMA, May 7, 2003 Vol 289, No