Author + information
- Received January 17, 2018
- Revision received April 3, 2018
- Accepted April 20, 2018
- Published online August 27, 2018.
- Laura P. Gelfman, MD, MPHa,b,∗ (, )
- Yolanda Barrón, MSc,
- Stanley Moore, BSd,
- Christopher M. Murtaugh, PhDc,
- Anuradha Lala, MDe,
- Melissa D. Aldridge, PhD, MBAa,b and
- Nathan E. Goldstein, MDa,b
- aBrookdale Department of Geriatrics and Palliative Medicine, Icahn School of Medicine at Mount Sinai, New York, New York
- bGeriatric Research Education and Clinical Center, James J. Peters VA Medical Center, Bronx, New York
- cCenter for Home Care Policy and Research, Visiting Nurse Service of New York, New York, New York
- dIndependent Contractor, Bonny Doon, California
- eDivisions of Cardiology and Population Health Science and Policy, Icahn School of Medicine at Mount Sinai, New York, New York
- ↵∗Address for correspondence:
Dr. Laura Gelfman, Department of Geriatrics and Palliative Medicine, Icahn School of Medicine at Mount Sinai, One Gustave L. Levy Place, Box 1070, New York, New York 10029.
Objectives This study sought to: 1) identify the predictors of hospice enrollment for patients with heart failure (HF); and 2) determine the impact of hospice enrollment on health care use.
Background Patients with HF rarely enroll in hospice. Little is known about how hospice affects this group’s health care use.
Methods Using a propensity score–matched sample of Medicare decedents with ≥2 HF discharges within 6 months, an Outcome and Assessment Information Set (OASIS) assessment, and subsequent death, we used Medicare administrative, claims, and patient assessment data to compare hospitalizations, intensive care unit stays, and emergency department visits for those beneficiaries who enrolled in hospice and those who did not.
Results The propensity score–matched sample included 3,067 beneficiaries in each group with a mean age of 82 years; 53% were female, and 15% were Black, Asian, or Hispanic. For objective 1, there were no differences in the characteristics, symptom burden, or functional status between groups that were associated with hospice enrollment. For objective 2, in the 6 months after the second HF discharge, the hospice group had significantly fewer emergency department visits (2.64 vs. 2.82; p = 0.04), hospital days (3.90 vs. 4.67; p < 0.001), and intensive care unit stays (1.25 vs. 1.51; p < 0.001); they were less likely to die in the hospital (3% vs. 56%; p < 0.001), and they had longer median survival (80 days vs. 71 days; log-rank test p = 0.004).
Conclusions Beneficiaries’ characteristics, including symptom burden and functional status, do not predict hospice enrollment. Those patients who enrolled in hospice used less health care, survived longer, and were less likely to die in the hospital. A tailored hospice model may be needed to increase enrollment and offer benefits to patients with HF.
Heart failure (HF) is a leading cause of death in the United States, and the number of individuals who are older than 65 years of age with advanced HF continues to rise (1). The incidence of HF approaches 10 per 1,000 population after age 65 years (2). Among patients who were hospitalized with HF between 2005 and 2009, patients had a 5-year risk for cardiovascular and HF admission of 80.4% and a median survival of 2.1 years (3). Approximately 80% of patients hospitalized with HF are older than 65 years of age, and up to 44% of older patients with advanced HF are readmitted within 6 months after discharge (4,5). Furthermore, the clinical HF trajectory typically is a gradual functional decline with intermittent acute deteriorations, which often require hospitalization and can be life threatening (6). Patients with HF also face increased complexity as their disease worsens, with the concurrent use of expensive and invasive medical interventions, along with the ongoing possibility of advanced medical therapies such as mechanical circulatory support and cardiac transplantation; although these therapies may prolong life, they also create barriers to hospice enrollment and greater difficulty in predicting mortality, as compared with other end-stage illnesses, such as cancer (7,8). Nonetheless, these patients face a high mortality; in a population cohort study, 5-year survival for HF stage D was 20% (9). Finally, patients with HF have a poor quality of life with untreated dyspnea, pain, fatigue, and depression (10,11). As these statistics demonstrate, better models of care are needed for patients with advanced HF and their caregivers.
Hospice is a model of care that could address the needs of patients with advanced HF and their caregivers. Specifically, hospice is a health care model for patients with a limited prognosis that focuses on symptom management, care coordination, and caregiver support, with the goal of enabling individuals to remain at home at the end of life. Use of hospice has risen dramatically in the past 2 decades from 10% to almost 50% of decedents, and hospices are caring for more patients with a primary diagnosis of dementia, HF, and chronic obstructive pulmonary disease (12,13). Furthermore, the use of hospice is considered to be an indicator of high-quality end-of-life care (14), and many studies have demonstrated its benefits for patients and families (15–23). Not only is hospice a key home care model for seriously ill patients, but also it is cost saving to the Medicare program (24,25).
Although the use of hospice by patients with HF has almost doubled in the past decade (26,27), the proportion of HF decedents who use hospice remains lower than that of other major diseases. Cardiovascular professional societies have called for the increased and earlier integration of hospice care for patients with advanced heart disease (7). Little is known about the characteristics of patients with HF who use hospice versus those who do not, and whether or not hospice affects health care use by patients with HF at the end of life. In a sample of Medicare beneficiaries with advanced HF, we: 1) examined the predictors of hospice enrollment after the second HF hospitalization within 6 months; and then 2) compared outcomes of health care use and survival for those patients who subsequently enrolled in hospice and those who did not.
We conducted a secondary analysis of data from a longitudinal, observational study that examined patterns of care in a cohort of patients with HF by using Medicare claims and patient assessment data (28). This existing dataset, obtained by our collaborators from the Visiting Nurse Service of New York Center, used 2009 and 2010 Medicare administrative claims and home health patient assessment data to identify all HF hospitalizations with discharge to home health care between July 1, 2009, and June 30, 2010 (28). The International Classification of Diseases-9th Revision-Clinical Modification codes used by the federal Chronic Conditions Data Warehouse were used to identify patients admitted with a principal diagnosis of HF (28). HF hospitalizations assigned to a surgical Medicare Severity Diagnosis Related Group were excluded because wound care, rather than HF management, was likely the primary reason for home health care (28).
Because the occurrence of multiple HF hospitalizations has been established as a marker of increased HF severity and risk of death (29), we selected a subsample of individuals with 2 or more HF hospitalizations within a 6-month period (July 1, 2009, to December 31, 2009) to represent a cohort of patients with advanced HF. To obtain data about functional status and symptoms, only beneficiaries with an Outcome and Assessment Information Set (OASIS) assessment by a home care clinician within 7 days of the second HF discharge (July 1, 2009, to December 31, 2009) were included for analysis, using follow-up data in the subsequent year (2010). Finally, only those patients who died within 6 months of the second HF discharge were included in the study group (Figure 1).
Sources of data
We constructed an analytic file of HF discharges of Medicare fee-for-service beneficiaries from 2009 to 2010 by linking data from the following Centers for Medicare & Medicaid Services administrative and claims files: 1) the Medicare Beneficiary Summary (Enrollment) file, which contains demographic and enrollment information about each beneficiary enrolled in Medicare; 2) the Medicare Provider Analysis and Review (MedPAR) file, which contains data on services provided by Medicare-certified inpatient hospitals and skilled nursing facilities (SNFs), including hospital admission and discharge date, discharge site (i.e., home to SNF), diagnoses, Diagnosis Related Groups codes, and intensive care unit (ICU) length of stay; 3) the Medicare Beneficiary Annual Summary (chronic condition indicators) file; 4) the Hospice Standard Analytic File, which includes claims data on hospice use; 5) the Medicare Home Health and Outpatient Standard Analytic Files, which contains home and outpatient claims data; 6) the Carrier File (Part B Standard Analytic File), which includes physician services information; and 7) the federally mandated OASIS collected by home care clinicians on admission to home care, discharge from home care, and resumption of care following an inpatient stay that interrupted the home care episode. OASIS includes information on the following: 1) patient home care diagnoses; 2) symptom assessment; and 3) activities of daily living (ADLs). Finally, to obtain regional measures of total Medicare spending at the end of life, we used publicly available data from the Dartmouth Atlas of Healthcare. The Institutional Review Boards of the Visiting Nurse Service of New York Center and the Icahn School of Medicine at Mount Sinai approved conduct of this study.
Our primary outcome was hospice enrollment after the second HF discharge. Other outcomes included health care use measures: number of inpatient days and stays, ICU days, SNF days, and emergency department (ED) visits; time to death after the second HF discharge; time to hospice enrollment after the second HF discharge; and location of death.
Exposure to hospice
In addition to serving as the primary outcome, hospice enrollment was the main exposure in comparing health care use, and time to and location of death, for patients who enrolled in hospice and those who did not.
We identified age, sex, and race from the Medicare administrative data. We measured comorbidity as a count of the number of conditions listed in Medicare claims for the 6 months before the index HF discharge. The specific comorbidities assessment included the Elixhauser score (30), as well as individual comorbidities including acute myocardial infarction, metastatic cancer, diabetes, Alzheimer’s dementia and related dementias, atrial fibrillation, depression, chronic kidney disease, arthritis, chronic lung disease, and obesity. The specific OASIS assessment included the sum of Katz’s ADLs (31), dyspnea assessment, pain assessment, use of oxygen treatment at home, and presence of impaired decision making. By linking beneficiaries’ ZIP codes to census data, we obtained the median family income in the beneficiary’s area of residence, as well as the total Medicare spending for each beneficiary’s ZIP code.
Predictors of hospice enrollment
To examine the predictors of hospice enrollment, we used chi-square or Student's t test where appropriate, to compare the relationship of functional status and symptom burden within 7 days of second HF discharge and use in the 6 months before the second HF discharge.
Propensity score matching
Because our study relies on observational data to assess how health care use and time until death after the second HF discharge differed by hospice enrollment, we conducted propensity score matching using the nearest neighbor method, also known as the greedy algorithm (32,33). Propensity score analysis controls for observed confounders that could influence group assignment by reducing the confounding covariates into a single variable, the propensity score. Matching methods balance the distributions of the observed covariates in the 2 groups, thus imitating what would have occurred in a randomized controlled trial. The propensity scores were created using the following variables: age, sex, race, income, whether patient lives alone, Medicaid enrollment, receipt of disability, census region, comorbidities, OASIS function and symptom assessment, and total Medicare spending. To account for regional health care use intensity, we also included regional total medical spending by matching the beneficiary ZIP code to hospital referral region level provided by the Dartmouth Atlas of Healthcare. Fifteen beneficiaries were excluded for missing ZIP codes. Next, we used a 3-group variable of total Medicare spending (lower quartile, 2 middle quartiles, and upper quartile) as an additional adjuster in the propensity score matching. All differences in beneficiaries’ characteristics disappeared after performing the propensity score matching (Table 1). All variables achieved balance with standardized bias <0.02. Because of an inability to match, 231 beneficiaries were excluded.
Health care use and survival
Using the propensity score–matched sample, we conducted bivariate analyses to compare health care use in the 30 days and the 6 months after second HF discharge of those patients who enrolled in hospice and those who did not. We estimated negative binomial (NB) regression models to compare use outcomes because all data were overdispersed or underdispersed on the basis of the likelihood ratio test; these outcomes were measured as counts, including the number of ED visits, inpatient days and admissions, ICU days, and SNF days, of those patients who did or did not enroll in hospice. We used 0-truncated NB models for outcomes with no values of 0, including number of all-cause hospital stays and days in the 6 months before the second HF discharge, because our exclusion criteria excluded all beneficiaries with 0 HF discharges. For use outcomes with a disproportionate number of 0s, we estimated 0-inflated NB models. These outcomes included ICU and SNF days, because a large proportion of patients did not use these services. Use outcomes for the propensity score–matched sample were compared using panel NB regression. If 0-truncated or 0-inflated models were preferred, cluster robust confidence intervals were used to compare use outcomes for the propensity score–matched sample. Vuong tests (34) were used to determine whether the estimation of a 0-inflated component was appropriate. We report adjusted marginal means and 95% confidence intervals.
Next, we compared survival between those patients who enrolled and who never enrolled in hospice using Kaplan-Meier curves and the log-rank test. Finally, we compared the location of death for those patients who did and did not enroll in hospice the location of death included hospice (home, inpatient, or at unknown setting), hospital, SNF or rehabilitation facility, or home health care. We used Stata software version 12.1 (StataCorp, College Station, Texas) to estimate these models.
We identified 234,301 Medicare fee-for-service beneficiaries with at least 1 home health claim and 1 HF hospital discharge in the 1-year interval between July 1, 2009, and June 30, 2010. Among them, 59,405 (25.4%) beneficiaries had 2 HF hospitalizations in a 6-month period, and of those patients, 22,893 (38.5%) had an OASIS assessment completed within 7 days of the second HF discharge. Finally, 7,810 (34.1%) died within 6 months of the second HF discharge (Figure 1).
Of the 7,810 patients who met our inclusion criteria, 3,298 (42.2%) enrolled in hospice, and 4,512 (57.8%) received usual care. The inclusion criteria were as follows: 1) 2 or more HF discharges in a 6-month period; 2) OASIS assessment within 7 days of second HF discharge; and 3) death within 6 months of second HF discharge. The propensity score–matched sample included 3,067 beneficiaries who enrolled in hospice and 3,067 who did not. Fifteen beneficiaries had missing ZIP code data, and 216 beneficiaries could not be matched in the propensity score; all 231 of these individuals were dropped (Figure 1).
The beneficiaries in the final cohort of propensity score–matched beneficiaries (n = 6,134) had a mean age of 82 years, 53% were female, and 15% were Black, Asian, or Hispanic (Table 1). Although those patients who were not matched by propensity score were generally similar to those who were matched, those who could not be matched by propensity score (n = 216) were more likely to be older (>85 years of age), be sicker (metastatic cancer, dementia, renal failure), and have worse functional status (more ADL or instrumental ADL dependencies), and they were less likely to have Medicaid or receive a disability entitlement.
Predictors of hospice enrollment
In the unmatched sample (n = 7,810), compared with those who did not enroll in hospice, beneficiaries who subsequently enrolled in hospice were more likely to be >85 years of age (45% vs. 34%; p < 0.001), white (86% vs. 78%; p < 0.001), have higher median household income ($53,193 vs. $50,761; p < 0.001), be sicker (metastatic cancer 9% vs. 6%; p < 0.001), have dementia (22% vs. 18%; p < 0.001), have dyspnea at rest or with minimal exertion (49% vs. 44%; p = 0.021), and have more functional dependencies (4.1 vs. 3.8; p < 0.001); they also were less likely to have Medicaid (19% vs. 29%; p < 0.001) and a disability entitlement (4% vs. 8%; p < 0.001) (Table 1).
In examining health care use in the 6 months before the second HF discharge, there were differences between the groups of beneficiaries who subsequently received and did not receive hospice. Compared with those patients who did not enroll in hospice, in the 6 months before the second HF discharge, those who subsequently enrolled in hospice had fewer ED visits (3.1 vs. 4.6; p < 0.001), all-cause inpatient admissions (1.70 vs. 1.88; p < 0.001), and all-cause inpatient days (12.4 vs. 14.0; p < 0.001), and there were no significant differences in the mean number of ICU days or SNF days for the unmatched sample (Table 2).
After propensity score matching, there were no significant predictors of hospice use, except that hospice patients still had fewer ED visits on average compared with beneficiaries who received usual care (2.38 vs. 2.59; p = 0.01) (Table 2).
Health care use after the second HF discharge: Hospice group versus usual care group
Table 3 reports the propensity score–matched and adjusted mean number of ED visits, all-cause hospital admissions (stays), mean number of all-cause hospital days, mean number of ICU days, and number of SNF days within 30 days after the second HF hospital discharge. Table 3 also presents the same outcomes after the second HF hospital discharge for the 6 months following hospital discharge. For the matched and adjusted sample within 30 days after the second HF hospital discharge, beneficiaries who enrolled in hospice had significantly fewer ED visits (2.64 vs. 2.82; p = 0.04) and spent fewer days on average in the hospital (3.90 vs. 4.67; p < 0.001) and in the ICU (1.25 vs. 1.51; p < 0.001) compared with those in usual care. There were no statistically significant differences in the number of inpatient stays within 30 days. When we examined similar outcomes for the 6 months after hospital discharge, beneficiaries who enrolled in hospice had on average significantly fewer ED visits, inpatient days, inpatient stays, ICU days, and SNF stays compared with beneficiaries who received usual care (Table 3).
Compared with those who received usual care, those who enrolled in hospice were less likely to die in the hospital (3% vs. 56%; p < 0.001) and less likely to die in an SNF (1% vs. 10%; p < 0.001). Of patients enrolled in hospice, 1,744 (57%) died with home hospice, 995 (32%) died in inpatient hospice, and 126 (4%) died in hospice in an unknown setting. The median number of days from second HF hospital discharge to enrollment in hospice was 51 (SD = 47 days). A total of 247 beneficiaries (8.1%) enrolled in hospice within 7 days after second HF hospital discharge, and 33.3% enrolled within 30 days after the second HF hospital discharge. The mean number of days from hospice enrollment to death was 22 ± 30 days. Nearly one-half (45%) died within 7 days. A total of 207 patients (6.7%) disenrolled from hospice during the 6-month study period, and of those, 95 (44.9%) disenrolled in the first 30 days after hospice enrollment. Those who ever enrolled in hospice had longer median survival compared with matched patients who received usual care (80 days vs. 71 days; log-rank test p = 0.004) (Figure 2).
Beneficiaries’ characteristics, including symptom burden and functional status, do not predict hospice enrollment. Those who enrolled in hospice used less health care, had longer survival, and were less likely to die in the hospital. In a sample of ill older adults with advanced HF, we examined predictors of hospice enrollment and compared health care use for those who enrolled in hospice and those who received usual care in the 6-month period following hospital discharge. Because 2 HF hospital discharges within a 6-month period comprise a marker of advanced HF, we observed a high symptom burden and poor functional status for patients with HF. Although those who enrolled in hospice after the second HF hospital discharge were older, more likely to be white, and less likely to have Medicaid and disability entitlement than those who received usual care, these trends are similar to national statistics about hospice enrollment across all diagnoses (35). Of note, in the unmatched sample, both those who enrolled in hospice and those who did not had a similar number of comorbidities and a similar symptom burden and functional status. Furthermore, health care use before the second HF discharge did not predict subsequent hospice enrollment because those patients who did not enroll in hospice had less health care use in the 6 months before the second HF discharge.
Compared with those who received usual care, beneficiaries who enrolled in hospice used less acute health care (including ED visits, inpatient stays and days, ICU days, and SNF days), had a longer survival, and were less likely to die in the hospital setting. Compared with similar patients not receiving hospice services, there is an estimated average cost savings of $2,300 per hospice beneficiary, which yields an overall savings of more than $3.5 billion a year for the 1.6 million patients served (36). These estimates do not account for the burdensome costs of informal caregiving. Nonetheless, these data demonstrate how hospice reduces formal health care use and, in turn, costs. These observations support several other studies demonstrating that patients who enroll in hospice experienced a longer survival (37), and that hospice enrollment reduced health care use and costs. Although many factors, including racial, socioeconomic, cultural, and religious factors, among others, may influence where individuals want to die, patients have stated that they prefer to die outside of the hospital (38). Of those who did not enroll in hospice, 56% died in the hospital compared with 3% of those enrolled. Although <7% of the hospice group disenrolled from hospice, nearly one-half of those who disenrolled from hospice did so in the first 30 days. Given the limitations of these claims data, we are unable to distinguish whether those disenrollees were discharged from hospice or revoked the hospice benefit.
Because our sample included only beneficiaries who received home health care before the 2 HF hospitalizations, this sample may have been sicker than other groups with advanced HF. In particular, these findings may not apply to patients referred to hospice who may be less sick (i.e., directly from the outpatient setting) and thus more likely spend more days enrolled in hospice. However, previous work suggests that patients and families with longer hospice lengths of stay report greater benefits from longer lengths of stay (39); therefore, our sample may underestimate these findings. Although the data are from 2009 and 2010, these results likely underestimate more current data because of the increased dissemination of hospice and palliative care into education, research, and clinical practice (7,8). Next, this sample included only Medicare beneficiaries who died; nonetheless, the mortality follow-back approach allowed us to characterize the use of resources for beneficiaries in the last 6 months of life. Finally, because the claims data did not include standard HF severity indicators, such as American College of Cardiology and American Heart Association HF stage or New York Heart Association HF functional class, advanced HF was represented by the history of 2 HF hospitalizations within a 6-month period as a surrogate for disease severity; despite the limitations of symptom reporting in claims data, such as OASIS, the high symptom burden and poor functional status of our sample demonstrates that our inclusion criteria effectively selected a high-risk group.
These data demonstrate that the marker of 2 HF hospital discharges within a 6-month period is an important indicator of high symptom burden and poor functional status. Yet, these beneficiary-related characteristics of multiple comorbidities, high symptom burden, poor functional status, and health care use are not predictive of subsequent hospice enrollment. Furthermore, patients who enroll in hospice use less health care, are more likely to die out of the hospital, and have longer survival. Given these benefits to patients with advanced HF, a more inclusive model for hospice is needed, such that patients are not required to agree to forego Medicare services aimed at curing disease or prolonging life. In addition, because of how difficult it can be to predict survival in these patients, changes to hospice should also include more flexibility around this requirement. Such a model of hospice for patients with HF that is tailored to the needs of this group would need to be adequately studied, with the goal to ensure that more people with advanced HF have access to these benefits.
COMPETENCY IN MEDICAL KNOWLEDGE: In older adults with advanced HF, patients’ characteristics, including symptom burden and functional status, do not predict hospice enrollment. Furthermore, those patients who enrolled in hospice used less health care, survived longer, and were less likely to die in the hospital. These findings suggest that a tailored hospice model may be needed to increase enrollment and offer benefits to patients with HF.
TRANSLATIONAL OUTLOOK: Future studies should evaluate the impact of a tailored hospice model for older adults with advanced HF that supports earlier enrollment in hospice to better support the needs of these patients.
Dr. Gelfman has received support from the National Institute on Aging (K23AG049930), The Mount Sinai Older Adults Independence Center (P30AG028741), and the National Palliative Care Research Center. The core dataset analyzed in this study was developed as part of a project supported by grant R01HS020257 from the Agency for Healthcare Research and Quality. The content is solely the responsibility of the authors and does not necessarily reflect the official view of the Agency for Healthcare Research and Quality or the National Institute on Aging or the U.S. Department of Veterans Affairs. The dataset for the zip code crosswalk was obtained from The Dartmouth Atlas, which is funded by the Robert Wood Johnson Foundation and the Dartmouth Clinical and Translational Science Institute, under award number UL1TR001086 from the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health (NIH). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- activity/activities of daily living
- emergency department
- heart failure
- intensive care unit
- negative binominal
- Outcome and Assessment Information Set
- skilled nursing facility
- Received January 17, 2018.
- Revision received April 3, 2018.
- Accepted April 20, 2018.
- Lloyd-Jones D.,
- Adams R.J.,
- Brown T.M.,
- et al.
- Shah K.S.,
- Xu H.,
- Matsouaka R.A.,
- et al.
- Heidenreich P.A.,
- Sahay A.,
- Kapoor J.R.,
- Pham M.X.,
- Massie B.
- Kavalieratos D.,
- Gelfman L.P.,
- Tycon L.E.,
- et al.
- Gelfman L.P.,
- Bakitas M.,
- Warner Stevenson L.,
- Kirkpatrick J.N.,
- Goldstein N.E.
- Ammar K.A.,
- Jacobsen S.J.,
- Mahoney D.W.,
- et al.
- National Hospice and Palliative Care Organization
- Earle C.C.,
- Park E.R.,
- Lai B.,
- Weeks J.C.,
- Ayanian J.Z.,
- Block S.
- Wright A.A.,
- Keating N.L.,
- Balboni T.A.,
- Matulonis U.A.,
- Block S.D.,
- Prigerson H.G.
- Bradley E.H.,
- Prigerson H.,
- Carlson M.D.,
- Cherlin E.,
- Johnson-Hurzeler R.,
- Kasl S.V.
- Kelley A.S.,
- Deb P.,
- Du Q.,
- Aldridge Carlson M.D.,
- Morrison R.S.
- Aldridge M.D.,
- Epstein A.J.,
- Brody A.A.,
- Lee E.J.,
- Morrison R.S.,
- Bradley E.H.
- Aldridge M.D.,
- Bradley E.H.
- Murtaugh C.M.,
- Deb P.,
- Zhu C.,
- et al.
- Mehta H.B.,
- Sura S.D.,
- Sharma M.,
- Johnson M.L.,
- Riall T.S.
- Long J.S.,
- Freese J.
- National Hospice and Palliative Care Organization