New Study Analyzes Geographic Disparity In End-Of-Life Care
Findings Challenge Assumptions Of Current Reform Efforts
HANOVER, N.H. -- The number of hospital beds in a community is the strongest determinant of whether a terminally ill patient dies in the hospital or at home, according to a new study published today in the Journal of the American Geriatrics Society (JAGS). The study, which analyzed the United States' geographic disparities in end-of-life care, found that neither patients' preferences nor the seriousness of their illnesses could explain the dramatic differences in care across the country.
The findings seriously challenge current efforts to reform end-of-life care. Most of the present initiatives rest on the assumption that patients, when armed with better information about their choices and encouraged to take a more active role in health care decision-making, can control the course of their own lives while dying. But the study results indicate that unless local health care resources and practice styles are adjusted to meet the needs of dying patients, patients' wishes likely will not prevail.
The research brought together data from the SUPPORT study (The Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments), and the Dartmouth Atlas of Health Care, both funded by the Robert Wood Johnson Foundation. The Dartmouth Atlas, which examined geographic variations in health care, found dramatic differences in end-of-life care.
SUPPORT is the largest clinical investigation ever conducted in the United States with hospitalized, seriously ill patients. In SUPPORT, patients with nine serious illnesses at five medical centers were provided with detailed information about their prognosis, as well as a highly trained nurse to facilitate communication and improve pain control. Unfortunately, this effort did not improve the patients' experiences. The JAGS study may help explain why the effort was ineffective.
"Sadly, although many people want to do something to avoid an impersonal death, without dignity or comfort, their fates often hinge on impersonal, even invisible, factors, such as the number of hospital beds in their community," said the study's first author, Robert S. Pritchard, M.D., who was on the faculty of Dartmouth Medical School and a member of the Veterans Affairs Medical Center, White River Junction, Vt., when the study was conducted. He has since joined the staff of Martha Jefferson Hospital in Charlottesville, Va.
Study co-author, Elliott S. Fisher, M.D., MPH, said, "More than anything else, the inner workings and resources of the local health care system determine how people's lives end. Even an advance directive generally will not be enough to overcome these forces." Fisher is an associate professor at Dartmouth Medical School and a member of the White River Junction VA Medical Center.
In the new study, researchers conducted two levels of analysis. First, they examined a subset of 479 SUPPORT patients who had been surveyed about their preferences regarding death and who later died. The researchers also conducted a national cross-sectional analysis of Medicare beneficiaries who died in 1992 and 1993.
First, researchers found that patients with any particular set of illnesses and other characteristics at one site were much more likely to die in the hospital than were those at some other sites. Preferences expressed by patients in the SUPPORT group had no effect on where those patients died. Differences in other patient characteristics did not explain the large variations across the study sites. Instead, when more local hospital beds were in use, more deaths happened in the hospitals, in both the SUPPORT and Medicare data.
Of the SUPPORT patients, 81 percent said they preferred to die at home. Nevertheless, 55 percent died in a hospital, the researchers found. The percent of patients who died in a hospital varied by more than twofold across the five study sites, with rates ranging from 29 to 66 percent. Among Medicare beneficiaries, 39 percent of all deaths during 1992 and 1993 occurred in a hospital, with rates ranging from 23 percent in Portland, Ore., and Ogden, Utah, to 54 percent in Newark, N.J.
In Medicare, the hospital day rate, the number of hospital days per 1,000 beneficiaries, explained 82 percent of the variance in place of death, according to the research findings. A decrease of just 1 hospital bed per 1,000 population was associated with a 3.8 percent reduction in the in-hospital death rate. Increased hospice spending was also linked with lower rates of in-hospital death.
"Faced with the challenges of managing severely ill and dying patients, what is easiest for the system to do is what gets done," said co-author, Joanne Lynn, M.D., of the Center to Improve Care of the Dying at George Washington University. "We cannot overcome these patterns by simply providing patients with better information and improving communication. We need to take a hard look at the ways that systems influence care for dying patients and adjust them when necessary, so that the usual patient gets good care, even if he or she does not speak out. We must respect the wishes of seriously ill people, make their deaths less painful, and allow them to live more meaningfully at the end."