Hospice Dis-Enrollment and Quality of Care at the End-of-Life

1. Hospice Dis-Enrollment and Quality of Care at the End-of-Life Melissa D.A. Carlson, Ph.D., M.B.A. Brookdale Department of Geriatrics & Adult Development Mount Sinai School of Medicine Academy Health Annual Meeting June 2008

2. Co-Authors • Jeph Herrin, Ph.D. • Qingling Du, M.S. • Andrew J. Epstein, Ph.D. • Emily Cherlin, M.S. • Sean Morrison, M.D. • Elizabeth H. Bradley, Ph.D.

3. Funding • National Cancer Institute (1R01CA116398-01A2) • Dr. Carlson is a Brookdale Leadership in Aging Fellow and recipient of an NINR Career Development Award (1K99NR010495-01) • Dr. Bradley is supported by the Patrick and Catherine Weldon Donaghue Medical Research Foundation Investigator Award (Grant #02-102) No disclosures or conflicts of interest

4. What do Older Adults With Serious Illness Experience? Not enough contact with MD: 78% Not enough emotional support (patient): 51% Not enough information about what to expect with the the dying process: 50% Not enough emotional support (family): 38% Not enough help with symptoms: 19% Although most people report wanting to die at home, less than 20% do so* Teno et al. JAMA 2004;291:88-93; Gallo, 2001

5. What is Hospice? • Multidisciplinary approach to care for patients suffering from serious and life-limiting illness • Covered by the Medicare Hospice Benefit • Patients are eligible if they have life expectancy of 6 months or less if disease follows its normal course • Willing to forego regular Medicare services focused on cure for primary illness • Hospice services include nurse and physician visits, pain management, counseling (spiritual or other), homemaker, respite care, and bereavement counseling • 75% of hospice care is provided in the home

6. Hospice Use is Increasing • 1.6 million hospice users in 2006 • 4/5 of these individuals were age 65+ • 1/3 of these individuals were age 85+ • 36% of all decedents in 2006 were under the care of a hospice program

7. Timing of Hospice Care Life Prolonging + Restorative Treatments Hospice PalliativeCare Disease Progression Bereavement Diagnosis Death

8. 10-15% of Patients Dis-Enroll from Hospice prior to Death • Dis-enrollment is associated with patient demographic and clinical factors: • Non-white1; Male 2; Younger age 1,2; Non-cancer diagnosis 1,2 • No difference by race2,3, age3 gender1,3, diagnosis3 • We know little about what happens to individuals once they dis-enroll 1Johnson 2008; 2Casarett, 2001; 3Taylor 2008

9. Reasons for Dis-enrollment • Patients initiated dis-enrollment: • Desire to pursue curative care • Dissatisfaction with hospice care • Transfer out of service area • Hospice initiated dis-enrollment • Unable to manage patients’ condition at home • Patient hospitalized or admitted to nursing home • Patient no longer eligible for the MHB (i.e., “fails to die in a timely manner”) • Important given regulatory pressure and audits of long-stay patients particularly those with non-cancer diagnoses (e.g., dementia)

10. Why do we care about dis-enrollment? • May be at high risk for claims-based indicators of quality of end of life care: • Hospitalization • ICU use • Emergency department use • Hospital death • Patient/family miss potentially beneficial services (bereavement counseling, supportive home care) • Unmet physical, psychological, spiritual needs

11. Specific Aims • Determine the association between hospice dis-enrollment and claims-based indicators of quality of care • Determine the association between hospice dis-enrollment and • Hospice characteristics (age, size, ownership, staffing levels) • Market characteristics (competition, region, urban/rural location)

12. Methods • Data: Linked SEER-Medicare claims data • Sample: Individuals who died with a primary diagnosis of cancer between 1998-2002 and who used hospice prior to death

13. Characteristics of the Sample • Patients (N=90,826) • Average age was 78.5 years • 86% White non-Hispanic • 52% Female • Hospices (N=1,384) • 77% more than 5 years old • 89% small (<50 patients w/cancer/year) • 63% non-profit

14. Poor Claims-Based Quality Outcomes for Patients Who Dis-Enroll P<0.001 for each comparison

15. Variation Across Hospices in Patient Dis-Enrollment Rate(Average=10.9%)

16. Factors Associated with Dis-Enrollment • Controlled for age, gender, comorbidity index value, home/inpatient hospice, staffing levels, urban/rural, region, and year • Generalized linear model accounting for the correlation of patient obs. within hospices

17. Limitations • Sample includes only patients with cancer • Quality indicators are those measurable with claims data; better measures addressing the needs of pateints who dis-enroll are needed

18. Implications • Dis-enrollment appears associated with poor quality care; better measures are needed • Higher dis-enrollment by minorities remains concerning • Variation across hospices in dis-enrollment potentially indicates differences in ability to meet the needs of some patient groups • Higher dis-enrollment in competitive markets could indicate a practice of “casting a wide net” • critical as the number of hospices continues to increase at a dramatic rate