Introduction. Anthony W. Ireland 1,2. Abstract

Transcription

1 Access to palliative care services during a terminal hospital episode reduces intervention rates and hospital costs: a database study of elderly patients dying in hospital, Anthony W. Ireland 1,2 1 Department of Veterans Affairs, Canberra, ACT, and 2 School of Public Health, University of Sydney, Sydney, New South Wales, Australia Key words palliative care, hospital, costs, intensive care, cancer. Correspondence Anthony W. Ireland, PO Box 860, St Ives, NSW 2075, Australia. tony.ireland@dva.gov.au Received 6 September 2016; accepted 1 February doi: /imj Abstract Background: The burden of healthcare costs for persons approaching death is of increasing concern. This study examines cost savings associated with access to palliative care () during a hospital episode ending in death for a large sample of elderly patients. Methods: A retrospective cohort study of administrative data for the Department of Veterans Affairs clients identified patient demographics, hospital characteristics, utilisation data and component costs for the hospital terminal episode for patients aged 70 years who died in hospital between July 2011 and June Differences between patients with and without access to were analysed with descriptive statistics and negative binomial regression models. Results: Access to service was reported for 33.2% of patients, 59.5% for those with a cancer diagnosis and 24.3% for other patients. Rates were significantly lower in private hospitals for all patient groups. For the complete sample, access was associated with significantly lower rates of admission into the intensive care unit (1.9% vs 10.6%, P < 0.001), fewer coded procedures and lower costs for hospital accommodation, medical and diagnostic services. Mean total cost for terminal episodes was $ for patients and $ for those with no recorded access (P < 0.001). All differences remained significant after adjustment for patient age, comorbidity and hospital type. Conclusion: In a hospital episode ending in death, access to services was associated with significantly lower rates of medical interventions and total hospital costs. Introduction Of the Australians who died in , died in hospitals. Among hospital deaths, (32%) were in designated palliative care () beds, and a further 8863 deaths (12%), mostly in acute wards, were coded as palliative. 2 The proportion of deaths occurring in hospitals is steadily decreasing, while the proportion of in-hospital deaths with status as identified in hospital databases has increased from 37% in Funding: None. Conflict of interest: None. Added information: The findings and conclusions of this study are those of the author and do not necessarily reflect those of the Department of Veterans Affairs. The author is a contracted medical adviser to the Commonwealth Government (Department of Veterans Affairs) and this study was performed as part of these contracted duties to 44% in These data reflect international trends. 4,5 In Australia, as in many other countries, healthcare costs are known to rise sharply in the final months of life, particularly for patients with cancer. 6 9 The greater part of these costs, as high as 90%, are incurred by hospital admissions. 7 9 There is also current recognition that expensive, cure-focused hospital interventions may be inappropriate for elderly patients with advanced and complex comorbidities. 10,11 Surveys of general populations are cited to suggest a strong majority preference for dying at home. 12 However, there is clear evidence that care preferences change towards admitted care among sick persons approaching death 13,14 and among older persons. 12,15 Lack of physical or family resources in the home, concern over adequacy of symptom relief and desire to avoid causing a burden on family members, as well as unexpected clinical events 549

2 Ireland in terminal disease are all reasons for seeking hospital or other formal care. 15 Among patients already enrolled in a service, 60% overall and 80% of those with a prior hospice admission stated a preference for terminal care in a hospice unit. 16 The collective evidence thus points to hospital units continuing to provide a substantial proportion of terminal phase care. Access to specialist in the hospital environment has demonstrated effectiveness in reducing the symptom burdens of dying persons, whether these be physical or psycho-spiritual There are also established benefits for carers and families both during the dying period and in bereavement. 20 services for terminal hospital patients have also been associated with substantial cost savings. In various Australian situations, the bed-day costs of hospice units are substantially lower than those for acute services. 21,22 Elsewhere, access to in acute units has been shown to reduce rates of referral to the intensive care unit (ICU), of medical/surgical procedures and of major diagnostic procedures, with commensurate cost reductions To date, there have been no equivalent studies of the Australian experience. The objective of this study is to describe the impact of services on the costs of hospital episodes that end in death. Patient and hospital variables that may in turn impact the rates of access to services, and/or to the dimensions of any resulting cost differences, will also be identified to the extent achievable from administrative data. This is the first report of such details for a large national cohort in Australia. Methods The sample for this retrospective cohort study was drawn from the Department of Veterans Affairs (DVA) hospital databases with respect of all clients who had died in hospitals across Australia between July 2011 and June The study describes hospital episodes that ended in death, identified by separation code 8, which are referred to herein as terminal episodes. Database items for public and private hospitals included patient age and gender, principal and secondary diagnoses, length of stay (LOS), ICU admissions, operative procedures, admission from residential aged care (RAC), itemised costs and geographic location of the hospital. Patients younger than 70 years were omitted from the final analysis because of an atypical age distribution and incomplete hospital records due to relatively high proportions of patients holding DVA White Cards, which fund hospital services for a limited range of conditions. Those who died with a terminal episode of less than 2 days were also excluded on the grounds that may not impact service delivery in a very short terminal episode. Records with outlier LOS values exceeding 180 days were also excluded as the proportion of any involvement in such episodes was unknown but probably small. was identified according to the preferred method of the Australian Institute of Health and Welfare (AIHW), namely the presence of the ICD-10-AM code (Z51.5) in the diagnosis string and/or care type code inclusive, which identifies units. 2 All patients identified by these means were defined in this study as having access to services; the nature, intensity or duration of any involvement in any specific episode were not definable. Patients were classed as being in acute care if the service type code was 1.0, with access identified through the code Z51.5. The terminal diagnosis was classified as malignant disease (cancer) if an appropriate code (ICD-10-AM C00- C99 inclusive) appeared as the principal or within the first five secondary diagnoses for the terminal episode. This study did not attempt to differentiate findings according to specific non-malignant diagnoses. Comorbidity was assessed using a modification of the Charlson comorbidity index. 26 Patients who had operative procedures were identified by the procedure codes inclusive and/or the listing of a theatre fee. Hospital costs were shown in the database as the charges and fees approved and paid by the DVA for accommodation, theatre and prostheses. For public hospitals, costs were determined by contracts between the DVA and the various State health administrations. Individual contracts were made with private hospital proprietors. Fees raised by private providers of medical, allied health, pathology, diagnostic imaging and other clinical services with respect to the hospital episode were also computed. No pharmaceutical data were included. Costs were expressed in Australian dollars: no corrections were made for the inflation of health costs over the study period. Statistical analyses Univariate analyses were conducted to compare demographic and hospital utilisation data for patients who did and did not have reported access. Student s t-test and Pearson s Chi-square test were applied for continuous and categorical variables respectively. As many of the continuous data items had very wide variances, multivariate analyses were performed using negative binomial regression. Predictor variables, drawn from the data items listed above were included in the regression model if P <0.25 in the univariate analyses and remained in the 550

3 Palliative care and hospital costs final models if P <0.05 after backwards elimination. Referent values were assigned arbitrarily. All analyses were performed using SAS 9.3 (SAS Institute Inc., Cary, NC, USA) or Excel 2010 (Microsoft Corporation, Redmond, WA, USA). Ethics approval was obtained from the DVA Human Research Ethics Committee in December Results There were recorded deaths of DVA clients during the study period, of which (36.9%) died in hospitals across Australia. This study analysed data with respect to of the hospital deaths after the exclusion of 1171 persons aged <70 years, 4902 with terminal LOS = 1 day and 41 persons with LOS > 180 days. Of this sample, (53.5%) were males. The mean age was 88.5 years for males (median 89 years) and 88.3 years (median 88 years) for females. Patients aged 90 years or older comprised 45.4% of the total (Table 1). The majority of deaths (54.5%) occurred in public hospitals for both males and females. A quarter of this elderly sample (and 30% of those aged 90 or older) were admitted from RAC facilities, with significantly higher proportions among females (30% vs 22%, P <0.001). Malignant disease was identified in 2430 patients (22.6%) in public hospitals and 2573 (28.7%) in private hospitals. For 1464 patients (7.4%), the terminal episode included an ICU admission, which exceeded 3 days for more than half of these instances. One or more surgical procedures were performed during the terminal episode in 13.9% of patients (Table 1). access was recorded for 6551 patients (33.2%). The access rate for all patients decreased with increasing age, from 39.6% for patients aged years to 30.7% for those aged 90 years, (P < 0.001), but this trend was not apparent when patients with and without a cancer diagnosis were assessed separately, as shown in Table 2. For these and all other subgroups listed in Table 2, rates of access were significantly lower in private hospitals, the overall rate Table 1 Patient characteristics and key service elements: patients who died in hospital Males (N = ) Females (N = 9169) All (N = ) N % N % N % Age group (years) State NSW/ACT Victoria Queensland SA/NT Tasmania Western Australia Charlson comorbidity score Intensive care >72 h Procedure Admit from RAC Age Cancer diagnosis Palliative care Cancer Other diagnoses Private hospital Patients aged 70 years, terminal episode length of stay (LOS) >1 <180 days. Percentage of cancer patients receiving palliative care (= 1703/2915 for males). RAC, residential aged care. 551

4 Ireland Table 2 Proportions of palliative care access in public and private hospitals by patient age, diagnostic group and location of hospital patients aged >70 years, Public hospitals (N = ) Private hospitals (N = 8964) Cancer (N = 2430) Other (N = 8313) Cancer (N = 2573) Other (N = 6391) Age group (years) All ages Remoteness Outer/remote Inner regional Metropolitan Admitted from RAC All values are percentages. RAC, residential aged care. being 24.2%, compared with 40.8% in public hospitals (P < 0.001). Of the 6551 episodes with access, 4008 (61.2%) were in designated units (care type 3). This represented 20% of all hospital deaths in the study population. Cancer was the assigned diagnosis for 54% of deaths in type 3 services. There were 2445 patients who died in an acute care service (care type 1) with access, comprising 17% of this large subgroup (29% of cancer patients and 14% of other patients). Overall, access was recorded for 2977 of 5003 patients with cancer (59.5%) and 3574 (24.3%) patients with other diagnoses (Table 1). was associated with lower mean values of LOS for terminal episodes among patients with non-cancer diagnoses but not for cancer patients. This applied in both public and private hospitals. In public hospitals, access was associated with a fivefold reduction in the likelihood of admission to the ICU and in the mean time per capita in the ICU (Table 3). Rates of operative procedures for public patients were a quarter of those for non- patients, with greater dimensions of difference for cancer patients. In private hospitals, there were parallel differences of a lesser but still significant degree (P < for all items). ICU admissions and coded procedures were significantly more frequent in private hospitals for all patient groups (Table 3). Multivariate analyses confirmed that was associated with a reduction in mean LOS, equivalent to 18% of baseline for non-cancer patients (P < 0.001), but there was no significant impact on patients with cancer (P = 0.76). Patient age, comorbidity scores and state (of hospital) were significant determinants of LOS for non-cancer patients but were not significant factors for cancer patients (data not tabled). In public hospitals, -associated reductions in mean total costs for a terminal episode were $6662 for cancer patients, $7477 for other patients and $6016 overall in unadjusted analyses. Corresponding values in private hospitals were $3812, $2908 and $3075. The overall figures represented a 28.8% reduction of the cost incurred by non- patients. Component costs for hospital accommodation, medical fees and diagnostic procedures for all groups were significantly lower for patients (Table 3). The -related cost reductions became substantially greater with decreasing patient age: mean reduction for all patients aged 90 years was $3656 and $ for patients aged years (Fig. 1). In multivariate models, increasing patient age beyond 80 years, state (Victoria, South Australia, Tasmania) admission for RAC and private hospitals were associated with relatively lower costs for all patients, as was the male gender for cancer patients only. Total costs increased with increasing comorbidity, and related reductions also increased but for non-cancer patients only. When adjusted for all significant factors in the model, was associated with cost reductions equivalent to 27.8% of the defined baseline values for cancer patients and 35.6% for other patients (Table 4). The negative binomial model had a dispersion of 0.69, and the deviance/degrees of freedom quotient was 1.12, indicating satisfactory goodness of fit. Patients who were admitted to the ICU incurred mean total costs of $36 610, increasing to $ if an operative procedure was also involved. access was associated with substantial cost reductions with respect to ICU admission or a procedure performed separately but not for patients who experienced both events. For the 83% of the entire sample who had neither ICU admissions nor procedures, the mean cost was $ with a - 552

5 Palliative care and hospital costs Table 3 Associations between palliative care access, length of stay and hospital costs for terminal episodes Deaths in public hospitals (N = ) Cancer patients Other patients All patients Item N = 1814 N = 616 = 2572 = 5741 N = 4386 N = 6357 Length of stay ** ICU admission (%) ** ** ** ICU h ** ** ** Procedures (%) ** ** ** Hospital costs ($) Episode total ** ** ** Accommodation * ** ** Medical fees ** ** ** Diagnostic costs ** ** ** Deaths in private hospitals N = 8964 Cancer patients Other patients All patients Item N = 1166 N = 1407 N = 1002 N = 5389 N = 2168 N = 6796 Length of stay ** ICU admission (%) ** ** ** ICU h ** * ** Procedures (%) ** ** ** Hospital costs ($) Episode total ** ** ** Accommodation * ** * Medical fees ** ** ** Diagnostic costs ** ** **, palliative care;, no palliative care; ICU, intensive care unit. $, Australian dollars at current values, *P < **P < related cost difference ($) Figure 1 Palliative care ()-related cost reductions in terminal episode: by patient age. ( ), Cancer patients; ( ), other patients. related reduction of $1369 (P < 0.001). These data are illustrated in Figure 2. Of 970 patients who died in services other than palliative or acute care, access was recorded for 101, most of whom had cancer. The heterogeneity within this subgroup negated further meaningful analyses. For patients, 51.6% had a hospital episode preceding and continuous with the terminal episode compared with 27.8% for non- patients (P < 0.001). The duration and costs of these preceding episodes were similar for both patient groups (Fig. 3). Pre-terminal ICU admission was more common (2.5% vs 1.7%, P < 0.001) among those who subsequently had access. In the terminal episodes, greater -related reductions in rates of ICU admissions, procedures and total costs were seen for patients who had immediately prior hospital episodes (data not tabled). Discussion Access to services during the terminal hospital episode was found for one-third of all hospital deaths in this elderly study population. Access rates were higher in public hospitals, for patients with a cancer diagnosis and for younger patients. Lower rates of access with increasing age and for patients admitted from aged care facilities reflected lower proportions of malignant disease in these subgroups. Lower rates of access in private hospitals were partly explained by the concentration of specialist 553

6 Ireland Table 4 Determinants of costs for terminal hospital episodes: patients dying in hospital Diagnosis Baseline Cancer (N = 5003) Other diagnoses (N = ) Added cost 95% CI P Added cost 95% CI P Gender (male) 2166 <0.001 Age group (years) <0.001 < Referent Referent (-4595,-3355) 4618 (-6063,-674) ( 3897, 918) 7096 ( 8212, 5908) ( 4577, 1577) 9275 ( , 8216) ( 3272, 2615) ( , 9277) State <0.001 NSW/ACT Referent Referent Victoria 2122 ( 3443, 708) 3839 ( 4595, 3355) Queensland 287 ( 1936, 2751) 265 ( 1477, 1010) SA/NT 1066 ( 3797, 2065) 4420 ( 5997, 2709) Tasmania 1354 ( 2684, 63) 1582 ( 2383, 752) Western Australia 1270 ( 756, 3489) 1107 ( 2302, 155) Charlson score 0.01 < Referent Referent ( 3871, ) 341 ( 1190, 538) ( 970, 3807) 2885 (1975, 3828) (655, 6125) 6506 ( ) RAC 4593 ( 6093, 3054) < ( 661, 5007) <0.001 Private hospital 4109 (5039, 3129) < ( 1470, 146) Palliative care 6282 ( 7074, 5449) < ( 9330, 8350) <0.001 Costs in Australian dollars at values current in Baseline = value for patient from NSW/ACT aged years, nil comorbidity, not from RAC, dying in public hospital with no palliative care. CI, confidence interval; RAC, admitted from residential aged care. Dollars ICU+Proc ICU Proc Neither Figure 2 Mean costs of terminal episodes by service provision; deaths in hospital ICU, intensive care unit; Proc, procedure;, no palliative care;, palliative care access; ( ), ; ( ),. (hospice) units in the public sector. was associated with significantly lower hospital costs in the order of $ for all patient groups, but most particularly for those with non-cancer diagnoses. This study was based on a large sample of approximately 11% of all deaths of Australians aged 70 years in the study years. The age distribution of the DVA population was skewed towards the extremes of old age (Table 1). The proportion of deaths of DVA clients that occurred in hospital (38.3%) was below that reported for the whole Australian population (50%) in the study years, as was the overall percentage of access. 3 These differences reflected the age distribution of the study sample. Approximately 60% of subjects aged 85 years would have been aged care residents at the time of their death and would mostly have died in place. 27 Cost savings attributable to services in hospitals have been extensively reported in international studies. 9,24,25,28,29 A large study based on six hospitals in the United States confirmed that a reduction in ICU admissions was a major contributor to the cost savings of the service. The proportional cost savings of the current study were at least equivalent to those found in America. 24,30 The present study was unable to confirm the reported finding that -associated cost savings were higher for patients with the highest comorbidity. 31 The analysis of hospital expenditure is a complex matter with a wide range of determinant variables. 9,24,25 In seeking to report a differential cost for services, location and type of hospital, patient age, diagnostic group, prior residency in RAC and comorbidity status have all been significant factors in this study. The known capacity of pre-terminal services in the community to prevent 554

7 Palliative care and hospital costs Cancer Other Cancer no Other no Figure 3 Mean length of stay and costs for hospital stay preceding and continuous with terminal episode; ( ), LOS (days); ( ), cost ($ 000)., palliative care. or abbreviate hospital admission for terminal phase care 23,32 could not be assessed from these data. All services billable to DVA were recorded for all patient groups. The concerns that unidentified costs of the service itself may discount reported cost benefits 33 are acknowledged. It is likely, however, that any such costs not included in contracts for public hospital services or not billed by private practitioners would be relatively small. Apart from the high percentage of deaths (45.6%) that occurred in private hospitals, terminal hospital care for DVA clients in this study was not considered to be atypical; the Department takes no role in the clinical content of hospital care. As -associated cost reductions increased with decreasing patient age, it is likely that average -associated savings would be greater in the (younger) general population. The coding of most data items used to identify and determinants of costs is episode-specific; this study has thus reported on the final hospital episode only. This is not to deny the high importance of pre-terminal events for determining and interpreting terminal-phase care. In , 57% of all terminal episodes in Australian hospitals involved transfers from episodes in the same or another hospital. 2 The details of the linked pre-terminal episodes in DVA data at least suggest that lower related costs in terminal episodes were not simply a continuation of preceding low-intensity care (Table 3). The data at this stage could not distinguish commenced de novo in the terminal episode from already in place at the start of these episodes. Future studies linking databases for DVA hospital and community-based services and with data from other sources may provide additional answers. Shorter terminal episodes and greater -related cost reductions have been identified for patients with diagnoses other than cancer. A probable reason lies in the irregular disease trajectory and more difficult prognostication for many chronic diseases, with belated recognition of the need for a shift to end-of-life care. 34 Despite increasing awareness of the needs of those dying from non-malignant conditions, cancer patients are still heavily overrepresented in hospital services. 2 Further examination of the interplay between these variables is beyond the scope of a database study, which cannot establish cause effect relationships. As already noted, these administrative data do not provide any information as to the duration, content or outcomes of service. In 2014, Palliative Care Australia noted the potential value of a comprehensive examination of -related costing data across Australia. 35 Notwithstanding the identified caveats, the findings of this study are a first step in this direction. Conclusion In a large sample, services of unspecified content have been identified for almost one-third of all hospital episodes ending in death. The total costs of these episodes were reduced by factors of 28% for cancer patients and 36% for other patients. Much lower rates of ICU admissions and operative procedures in recipients, more evident for cancer patients, were the chief drivers of cost reductions. Research that can identify the onset and content of services in hospitals will be required to confirm and expand the findings of this report. Acknowledgements Officers of the Data Analysis and Nominal Rolls unit within the Department of Veterans Affairs provided support in extracting the datasets. References 1 Australian Bureau of Statistics. Deaths Australia ABS Cat. No Canberra: ABS; Australian Institute of Health and Welfare. Australian Hospital Statistics Health Services Series No. 60. Cat. No Canberra: AIHW; Australian Institute of Health and Welfare. Palliative Care Services in Australia Canberra: AIHW; Teno JM, Gozalo PI, Bynnra JPW, Leland NF, Miller SC, Mordent NE et al. Change in end-of-life care for Medicare beneficiaries. Site of death, place of care and health care transitions in 2000, 2005 and JAMA 2013; 309: Houttekier D, Cohen J, Surkyn J, Deliens L. Study of recent and future trends in place of death in Belgium using death certificate data: a shift from hospitals to care homes. BMC Public Health 2011; 11:

8 Ireland 6 Langton JM, Reeve R, Srasuebkul P, Haas M, Viney R, Currow D et al. Health service use and costs in the last 6 months of life in elderly decedents with a history of cancer: a comprehensive analysis from a health payer perspective. Br J Cancer 2016; 114: Tran B, Falster MO, Gresi F, Jorm L. Relationships between use of general practitioners and healthcare costs at the end of life: a data linkage study in NSW, Australia. BMJ Open 2016; 6: e Riley GF, Lubitz JD. Long-term trends in Medicare payments in the last year of life. Health Serv Res 2010; 45: Simoens S, Kutten B, Keirse E, Berghe PV, Beguin C, Desmedt M et al. The costs of treating terminal patients. J Pain Symptom Manage 2010; 40: Queensland Government. End of life care in Queensland-admission to acute hospitals near the end of life. Health Statistics Branch, Queensland Health. August Katelaris A. Time to rethink end of life care. Med J Aust 2011; 194: Foreman L, Hunt R, Luke C, Roder D. Factors predictive of preferred place of death in the general population of South Australia. Palliat Med 2006; 20: Agar M, Currow D, Shelby-James TM, Plummer J, Sanderson C, Abernethy AP. Preference for place of care and place of death in palliative care: are these the same questions? Palliat Med 2008; 22: Janssen DJA, Spruit MA, Schols JMGA, Wouters EFM. Dynamic preferences for site of death among patients with advanced chronic obstructive pulmonary disease, chronic heart failure, or chronic renal failure. J Pain Symptom Manage 2013; 46: Gott M, Seymour J, Bellamy G, Clark D, Ahmedzai S. Older people s views about home as a place of care at the end of life. Palliat Med 2004; 18: Arnold E, Finucane AM, Oxenham D. Preferred place of death for patients referred to a specialist palliative care service. BMJ Support Palliat Care 2015; 5: Casarett D, Pichard A, Bailey FA, Ritchie C, Furman C, Rosenfeld K et al. Do Palliative care consultations improve patient outcomes? J Am Geriatr Soc 2008; 56: Le B, Watt J. Care of the dying in Australia s busiest hospital: benefits of palliative care consultations and methods to enhance access. J Palliat Med 2010; 168: Higginson IJ, Evans CJ. What is the evidence that palliative care teams improve outcomes for cancer patients and their families? Cancer J 2010; 16: Hudson PL, Remedios C, Thomas C. A systematic review of psychosocial interventions for family carers of palliative patients. BMC Palliat Care 2010; 9: Queensland Parliament. Palliative care and community care in Queensland: toward person-centred care. Report No. 22. Health and Community Services Committee, May Independent Hospital Pricing Authority. National Hospital Cost Data Collection: Australian Public Hospitals Cost Report , Round Gomez-Batiste X, Caja C, Espinosa J, Bullich I, Martínez-Muñoz M, Portasales J et al. The Catalonia World Health Organisation demonstration project for palliative care implementation: quantitative and qualitative results at 20 years. J Pain Symptom Manage 2012; 43: Morrison RS, Penrod JD, Cassel JB, Caust-Ellenbogen M, Litke A, Spragens L et al. Cost savings associated with US hospital palliative care consultation programs. Arch Intern Med 2008; 168: Smith TJ, Coyne P, Cassel B, Penberthy L, Hopson A, Hager MA. A high volume specialist palliative care unit and team may reduce in-hospital end of life care cost. J Palliat Med 2003; 6: Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method for classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40: Australian Institute of Health and Welfare Aged care data cubes: residential aged care and aged care packages in the community (Separations from residential aged care 1 July June 2012) [cited 2016 Sep]. Available from URL: aihw.gov.au/aged-care-data-cubes/ 28 Ellershaw JE, Peat SJ, Boys LC. Assessing the effectiveness of a hospital palliative care team. Palliat Med 1995; 9: Lo JC. The impacts of hospices on health care expenditures the case of Taiwan. Soc Sci Med 2002; 54: Gade G, Venohr I, Conner D, McGrady K, Beane J, Richardson R et al. Impact of an inpatient palliative care team: a randomized controlled trial. J Palliat Med 2008; 11: May P, Garrido MM, Cassel JB, Kelley AS, Meier DE, Normand C et al. Palliative care teams cost saving effect is larger for cancer patients with higher number of comorbidities. Health Aff 2016; 35: Siew H, Barbera L, Howell D, Dy S. Using more end-of-life homecare services is associated with using fewer acute care services: a population-based cohort study. Med Care 2010; 48: Brett AS, May PD, Garrido MM, Cassel JB, Kelley AS, Meier DE, Normand C, et al. Does palliative care consultation lower hospital costs? NEJM Journal Watch, September 10, Lynn J, Adamson DM. Living well at the end of life. Adapting health care to serious chronic illness in old age (Volume 137 of White Paper). Santa Monica, CA: Rand Corporation; Palliative Care Australia. Submission to the National Commission of Audit, January