VA Hospice and Palliative Care: Identifying Veterans at High Risk of Mortality

1. VA Hospice and Palliative Care: Identifying Veterans at High Risk of Mortality Ann Hendricks PhD, Lynn Wolfsfeld MPP Health Care Financing & Economics (HCFE) Boston VA Healthcare System, HSR&D April 23, 2008 Email: Ann.Hendricks@va.gov Lynn.Wolfsfeld@va.gov

2. Topics • Background on VA Hospice and Palliative Care • Overview of RRP • Analysis of Inpatients (Methods, Results, Conclusions) • Next Steps

3. VA HPC Initiatives • Establishment of National VA Office of Hospice and Palliative Care (2004) • Fellowship programs • Hospice-veteran partnership programs • Directive to establish palliative care consult teams at all VAMCs (2003)

4. Number of Palliative Care Consults Vary Across VISNs, FY2006-2007 There was an 11% national increase in the number of palliative care consults reported in FY07 (23,240) as compared to FY06 (20,943) for the 127 and 126 facilities reporting respectively. From: Hospice and Palliative Care Fiscal Year 2007 Status Report

5. Proportion of Inpatients Who Died who had Palliative Care Consults, FY2006-2007 For VA overall, the percentage of inpatient deaths with an associated palliative care consult increased from 42 to 47 percent for FY07. The average number of days between palliative care consult and death also increased from 45 to 47 days (with the median increasing from 34 to 37) from FY06 to FY07. The growth in these areas indicates greater and earlier involvement of palliative care teams with veterans approaching death while the variability among VISNs reflects the varying degrees of palliative care integration. From: Hospice and Palliative Care Fiscal Year 2007 Status Report

6. FY05 Strategic Initiative* • Improving access to hospice and palliative care in inpatient and outpatient settings • Exploration of automated case-finding techniques • 81% of facilities had no automated case finding method to identify veterans appropriate for HPC *From HPC FY 2005 Status Report

7. Project Objectives • To work with an expert panel to identify diagnoses and/or events in inpatient, outpatient and long term care settings that could indicated a referral to the hospice and palliative care team • To create computer algorithms for the indicators using data elements available in the various national VA databases • To determine the prevalence of the indicators • To test the final indicators agreed on by the expert panel by merging patients identified with the indicators with mortality data to see how predictive the indicators are

8. Expert Panel • National Director – Hospice and Palliative Care • Hospital Administrators (2) • Hospice and Palliative Care Specialists (2 VA – 1 non-VA) • National Chief – Hematology/Oncology • ICU Intensivist (1)

9. Mission Statement “To develop a practical tool which identifies veterans at substantial risk for needing specialized end-of-life care, often including palliative care and/or hospice services.”

10. Criteria for Inclusion in Case Finding Metric • Low-hanging fruit • Predicted probability of 50% or more of dying within a year • Cancers with poor prognoses • Multifaceted approach • Across settings and conditions (Inpatient, outpatient, nursing homes) • Patients close to death (days, months+) • ICU Events and Conditions • Chronic conditions

11. In-depth analysis of conditions meeting first criteria for inclusion • Low-hanging fruit • Predicted probability of 50% or more of dying within a year • Cancers with poor prognoses

12. Methods • VA NPCD Data FY 2001-2005 • VA Vital Statistics File (Mortality through March 2006) • Population – inpatients with cancer diagnoses • Index Date – first appearance of diagnosis (in VA) after 12 months with no care for that diagnosis (in VA)

13. Specifications (Populations) CANCERS ICD-9 CODES • Head/neck 141-148 • Trachea/bronchus/lung 162 • Prostate 185 • Colon 183 • Liver 155 • Pancreatic 157 • Esophageal 150 • Lymphomas 200-202 • Leukemias (acute) 204.0, 205.0, 206.0, 207.2, 207.7, 208.0 • Melanoma 172 • CNS 191 • All other cancers All remaining ICD-9 codes140-239

14. Regression (SAS – LifeReg) • Parametric Accelerated Failure Time Model • Allows for right censoring • Dependent Variable • Number of months survived • Independent Variables • Age • Gender • Advanced Disease • Separate model for each condition

15. Indications of Advanced Disease • ICD-9 196 = Secondary and unspecified malignant neoplasm of lymph nodes • ICD-9 197 = Secondary malignant neoplasm of respiratory and digestive systems • ICD-9 198 = Secondary malignant neoplasm of other specified sites for example (kidney, brain, skin, bone...)

16. ILLUSTRATIVE SURVIVAL CURVES AND MEDIANS

17. ILLUSTRATIVE SURVIVAL CURVES AND MEDIANS

18. THRESHOLD FOR INCLUSION IN CASE FINDING METRIC • Predicted probability of 50% or more of dying within 12 months • Equivalent to median predicted months survived <= 12

19. MEDIAN PREDICTED MONTHS OF SURVIVAL (FROM INDEX DATE), HOSPITAL INPATIENTS * ICD-9 196 = Secondary and unspecified malignant neoplasm of lymph nodes; ICD-9 197 = Secondary malignant neoplasm of respiratory and digestive systems; ICD-9 198 = Secondary malignant neoplasm of other specified sites for example (kidney, brain, skin, bone...)

20. MEDIAN PREDICTED MONTHS OF SURVIVAL (FROM INDEX DATE), WITHOUT INDICATION OF ADVANCED DISEASE, HOSPITAL INPATIENTS, BY AGE

21. MEDIAN PREDICTED MONTHS OF SURVIVAL (FROM INDEX DATE), WITH INDICATION OF ADVANCED DISEASE, HOSPITAL INPATIENTS, BY AGE

22. PRELIMINARY RECOMMENDATIONS FOR CONDITIONS TO INCLUDE IN A CASE FINDING METRIC FOR CANCER INPATIENTS – ACUTE SETTING

23. Next Steps • Assess V66.7 and TS96 codes • Refine current analysis - additional look at cancers – outpatients, co-morbidities, LOS • Additional analyses – chronic conditions, nursing home patients, outpatients, ICU patients, functional status • Implementation