Fit for the future: International comparisons in end-of-life care and what we can learn from them. Joachim Cohen

Transcription

1 Fit for the future: International comparisons in end-of-life care and what we can learn from them Joachim Cohen

2 What can we learn from the FIFA ranking?

3 What does it tell us? Is it valid? Is it important to us? How do we explain it? (Why?) Are differences acceptable? Can we learn from it / adapt?

4 What does it tell us? Is it valid?

5 Basic end-of-life healthcare environment (20%) Availability of end-of-life care (25%) Cost of end-of-life care (15%) Quality of end-of-life care (40%)

6 What does it tell us? Is it valid? Is it important to us? How do we explain it? (Why?) Are differences acceptable? Can we learn from it / adapt?

7 Some essentials about international comparative research

8 Three principal rationales for international comparative research 1. Learning about (describing) 2. Learning why (explaining) 3. Lessons learned from (identifying best practice)

9 Concepts cannot be separated from context Do concepts differ between countries in terms of: Conceptual, functional and semantic equivalence? Linguistic equivalence? Measurement equivalence

10

11 Mixed methods approaches in international research provide most enriching insights Cacace et al. Health Policy, 2013

12 Key points 1. characteristics or circumstances of death and dying determined more by country than by patient characteristics Large variations in: place of death place of care Hospital expenditures use of services 2. country-specific priorities in terms of allocation and quality assurance 3. healthcare organisational choices in terms of end-of-life care influence EOL patterns 4. An international comparative research agenda for EOLC is needed

13 Key points 1. characteristics or circumstances of death and dying determined more by country than by patient characteristics Large variations in: place of death place of care Hospital expenditures use of services 2. country-specific priorities in terms of allocation and quality assurance 3. healthcare organisational choices in terms of end-of-life care influence EOL patterns 4. An international comparative research agenda for EOLC is needed

14 Variation in place of death

15 International Place of Death (IPoD) Study Total: 5,570,065 deaths

16 population in need of palliative care by Rosenwax, McNamara et al. Underlying cause of death: Cancer Heart failure Renal failure Liver failure Chronic obstructive pulmonary disease Diseases of the nervous system HIV/AIDS

17 Population in need of palliative care Total: 2,220,997 deaths

18 Population in need of palliative care IT ES FR BE NL ENG WAL CZ HU NZ Western Australia: 50% McNamara et al, JPSM, 2006 Cancer Organ failure CA US KR MX Diseases of the nervos system HIV/AIDS

19 Large cross-national variation in place of death (N= 2,220,997) IT ES FR BE NL ENG WAL CZ HU NZ CA US MX KR Home Hospital Nursing home PC institution Other institution Other

20 Australia: few over 65 die at home Grattan report (2014)

21 Population dying of cancer Total: 1,355,910 deaths

22 Large cross-national variation in place of death (N= 1,355,910) IT ES FR BE NL ENG WAL CZ Home Hospital Nursing home NZ PC institutions Others CA US MX KR

23 In most countries cancer patient more likely to die at home IT ES FR BE NL ENG WAL CZ NZ CA US MX KR

24 Variation in place of care

25 International Consortium for End-of- Life Research Study Claims and registry data (2010 data) Total: 447,193 cancer deaths

26 Variation in hospital admission rates in last month N= 447,193 % with at least 1 admission last 180 days last 30 days BE CA ENG DE NL NO USA

27 4 3.5 Variation in mean hospital admissions and nr days spent in hospital in last month mean nr of admissions and mean nr of days N= 447,193 last 180 days last 30 days BE CA ENG DE NL NO USA

28 Variation in mean hospital admissions and nr days spent in hospital in last month mean nr of admissions and mean nr of days N= 447,193 last 180 days last 30 days BE CA ENG DE NL NO USA 0

29 45 40 Variation in ICU admissions and nr days spent there % with an admissions and mean nr of days N= 447,193 last 180 days last 30 days No data No data BE CA ENG DE NL NO USA

30 Variation in hospital expenditures

31 Resource Utilization and Hospital Expenditures in last 30-days of Life Average nr of days in hospital Hospital expenditures (in health specific PPP US$)

32 Variation in use of services

33 45 Variation in use of chemotherapy in final months N= 447,193 % with at least one chemotherapy episode last 180 days last 30 days No data BE CA ENG DE NL NO USA

34 So what?

35 Not useful because of the obvious differences in the health care systems, reimbursements and cultural attitudes

36 What does it tell us? Differences in how countries manage end-of-life care hospital-centric vs out-of hospital centric eg in cancer choices re: specialist palliative care services Country-specific priorities in terms of allocation and quality assurrance

37 Attention to end-of-life care in hospital: IT ES FR BE NL ENG WAL CZ HU Home Hospital NZ CA US MX KR Nursing home PC institution Other institution

38 Attention to home and care home as settings of end-of-life care : IT ES FR BE NL ENG WAL CZ HU Home Hospital NZ CA US MX KR Nursing home PC institution Other institution

39 What does it tell us? Differences in how countries manage end-of-life care hospital-centric vs out-of hospital centric eg in cancer Country-specific priorities in terms of allocation of quality assurrance Differences in spending Quality of EOLC issues benchmarking

40 Is it valid to us? Conceptual equivalence? eg hospital vs nursing home Comparable populations and methods Limited information context and contingencies

41 Is it important to us Yes quality of care rational use of resources planning of care (monitoring of needs and services within population)

42 How do we explain the differences Addressing the why question

43 Variation in home death only partly explained by clinical and sociodemographic patient characteristics and health care availability IT FR ES NL CZ ENG WAL CA US MX KR

44 Variation in home death only partly explained by clinical and sociodemographic patient characteristics and health care availability IT FR ES NL CZ ENG WAL CA US MX KR + cause of death, age, sex, marital status

45 Variation in home death only partly explained by clinical and sociodemographic patient characteristics and health care availability IT FR ES NL CZ ENG WAL CA US MX KR + availability of hospital beds, LTC beds, GPs

46 Cacace et al. Health Policy, 2013

47 In England: rise in hospital deaths followed by decrease Percentage of cancer deaths by place of death in England ( ) Could this be the effect of the End of Life Care Programme? Gao et al PLoS Med 2013

48 TRENDS IN PLACE OF DEATH IN BELGIUM All deaths % 50.0% 55.1% 51.7% 40.0% 30.0% 20.0% 10.0% 23.0% 22.5% 18.3% 22.6% Home Hospital Nursing home 0.0% Houttekier et al BMC Pub Health 2011

49 DECREASING PROPORTION DIES IN HOSPITAL IN BELGIUM 80.0% Proportion dying in hospital by living arrangement 70.0% 60.0% 68.6% 68.2% 63.4% 61.7% 50.0% 40.0% 30.0% 20.0% 31.0% 21.5% Single Multi-person household Nursing home 10.0%.0% Houttekier et al BMC Pub Health 2011

50 INCREASING PROPORTION DIES IN NURSING HOME IN BELGIUM Proportion dying in nursing homes by living arrangement 80.0% 75.3% 70.0% 66.8% 60.0% 50.0% 40.0% 30.0% Single Multi-person household Nursing home 20.0% 10.0%.0% 8.9% 5.7% 6.6% 3.7% Houttekier et al BMC Pub Health 2011

51 BELGIUM HAD A POLICY OF CONVERSION OF RESIDENTIAL TO SKILLED NURSING BEDS IN LONG-TERM CARE SETTINGS Hospital beds/ Residential Beds in Care Homes/ y Skilled Nursing Beds in Care Homes/ y Houttekier et al BMC Pub Health 2011

52 Unadjusted Odds Ratios of Nursing Home Relative to Hospital Death OR Year of Death Houttekier et al BMC Pub Health 2011

53 TRENDS IN NURSING HOME DEATH Adjusted Odds Ratios of Nursing Home Relative to Hospital Death AOR Year of Death EAPC YIA Odds ratios adjusted for age, sex, educational attainment, urbanization level, underlying cause of death, available residential beds and skilled nursing beds in nursing homes, and available hospital beds. Houttekier et al BMC Pub Health

54 US SAW REVERSAL OF TRENDS IN HOSPITAL DEATHS James Flory et al. Health Aff 2004;23: Flory et al Health Aff 2004 Teno et al JAMA 2013 Could this be the effect of policy changes?

55 VARIOUS FACTORS RESPONSIBLE FOR THE COUNTRY DIFFERENCES: Effects of specific end-of-life care policies Past choices regarding settings of end-of-life care (cancer vs non-cancer) Wider societal factors and historical contingencies

56 Perceptions about avoidability of a terminal hospitalization differ Country %in hospital % avoidable England 48% 7% Belgium 51% 14% New Zealand 28% 22% Netherlands 25% 24% Different studies

57 Are the differences acceptable? Yes: No: If preferences different No good empirical indications If QoL QoD is guaranteed despite differences hard to tell whether that s the case Too large to be logical Contingent but not arbitrary

58 Yes: Can countries learn from each other? Valuable insights from looking across countries Understanding similarity or specificity of problems Understanding policy development, ways to address problems, opportunities and constraints BUT: need other type of information Benchmarking Explain how and why policy measures or strategies are effective Context specificity key question: Under what circumstances and to what extent will a programme that works in country A also work in country B?

59 Fit for the future? Large-scale cross national comparison is only a first step Need for a research agenda based on learning, explaining and understanding

60 No too much omphaloskepsis

61 Key points 1. characteristics or circumstances of death and dying determined more by country than by patient characteristics Large variations in: place of death place of care Hospital expenditures use of services 2. country-specific priorities in terms of allocation and quality assurance 3. healthcare organisational choices in terms of end-of-life care influence EOL patterns 4. An international comparative research agenda for EOLC is needed