Dr. Julie BynumThere’s been a great deal of policy and public attention to hospice and hospice care recently, including the bankruptcy and closure of a large hospice in San Diego over inappropriately long stays and possible overcharging of the Medicare program. But I think this is the wrong story.
The effectiveness of hospice in meeting the needs of dying people and their loved ones is really challenged by stays that are far too short and a health care system that still does not support an orderly transition into the end of our lives.
I’ve written about this issue before (Throwing Out the Baby with the Bathwater? Part I and Part II) in 2011, but a new paper by the outstanding long-term care research team at Brown University, featuring Drs. Joan Teno and Vincent Mor and colleagues from Dartmouth, including Beeson program alum Dr. Julie Bynum, has sharpened the issue of appropriate hospice use and clarified where the problems lie.
Using the more and more current Medicare data sets that are increasingly available, they looked at the billing data of a huge 20 percent sample of Medicare beneficiaries who died in 2000, 2005, and 2009 to observe the patterns of care transitions and service utilization, including hospice use. To focus their lens, they also used the billing data to identify beneficiaries who had cancer, chronic obstructive pulmonary disease (COPD), or dementia in their last months. These conditions correspond to some of the most common trajectories of death: cancer—the inspiration for hospice care, which at the end still leads to relatively quick and predictable decline and death; COPD—a chronic disease with a longer period of decline but extremely burdensome symptoms (e.g., shortness of breath) combined with an unpredictable pattern of exacerbations and recoveries before death; and dementia—a disease characterized by a slower but steadier decline and very high levels of caregiver burden and loss of quality of life.
The first finding they report is despite the growth in use of hospice (overall from 21.6 percent of Medicare decedents 66 years of older in 2000 to 42.2 percent of decedents in 2009), the median number of hospice days of care in the last 30 days of life is still zero—yes, zero. Most people still don’t get hospice at all. And even at the 75th percentile, the number of hospice days was nine—which means that only 25 percent of those who died in 2009 received more than nine days of hospice care.
Focusing in on the specific conditions, the median (50th percentile) and upper 75th percentile number of days of hospice care for cancer was three and 19; for COPD, zero and six days; and for dementia, one and 13 days. This analysis, of course, combines people who don’t get into hospice at all along with the very short stays of those who do, but it tells us something about our real problems with end of life care even with conditions that are relatively slow and predictable killers.
As I’ve argued before, very short hospice stays undermine the effectiveness of the model as they deprive people and their families time to form relationships with the new healthcare team in their lives, time to implement care plans, and time to adjust to impending death. In 2011, the median time in hospice for those entering the program was only 16 days—and it clearly has not improved.
Drilling down into the course of care of those who do get hospice, the results show that there has been very little change in the ratio (almost 25 percent) of people with very short hospice stays (three or fewer days) to the total population getting hospice care. In 2000, 4.6 percent of all decedents had a short stay out of the 21.6 percent who had any hospice. In 2009, 9.8 percent had a short stay out of the 42.2 percent who got any hospice. In fact, the ratio of short stays to any use gets worse over time, with 23 percent of total hospice stays in 2009 being three days or less as compared to 21 percent in 2000.
So, despite our increased understanding of patterns of death, what should be increased prognostic ability for these diseases, and clear public interest in dying at home and with less heroic measures, people who get into hospice are still getting there way too late.
But the research reveals some other serious disconnects in care. For those who died in 2009, care was if anything more fragmented and chaotic in their last 30 days than it was for those who died in 2000—with transitions in site of care rising from around two in the last 30 days to three; use of ICU rising from 24.3 percent of cases to 29.2 percent overall (and similar growth in its use for the three focal conditions, even dementia); and the rate of having three or more hospitalizations in the last 90 days of life increasing from 10.3 percent of decedents to 11.5 percent. Only in the case of dementia does the frequency of these multiple hospitalizations decline, falling to 10.7 percent in 2009 from 12 percent in 2000.
While we don’t know from these data how well these patterns of care match specific patient and family preferences, I think the pattern of change is quite damning. I don’t think there is any conceivable way that people could want more stressful and chaotic care and ultimately ineffective care in 2009 than in 2000. It looks to me that while hospice has grown, it has not been implemented effectively. And it sure looks like providers of acute care have not learned anything about the limits of their services—or at least they are not sharing it in a timely way with patients.
There may be problems with end of life care, but it isn’t that people are getting too much hospice care.