End-of-life care has been fingered as a major culprit in high and inappropriate health spending. Research carried in Health Affairs looks at variation in end-of-life care for cancer patients and analyzes factors which may affect that variation. (HA Article) The authors used patient surveys, physician surveys and administrative data to examine spending in the last 30 days of life for patients who died from lung or colorectal cancer in 26 hospital referral regions. Some of the factors examined were patient demographics, patient attitudes toward care, physician attitudes toward care, and health resource levels in the area. About 1130 decedents were included, three-quarters of whom had lung cancer. 57% were men, 80% were Caucasian, the average age was about 76 years at death and average incomes were less that $40,000. Average spending in that last month of life was $13,663; with a range from $10,131 in the lowest spending quintile of HRRs to $19,318 in the highest spending one. The higher spending HRRs tended to be more populous, have larger shares of non-white residents, more physicians per 10,000 people, a lower percentage of doctors who practice primary care, and have fewer hospital beds and hospices.
Physician beliefs appeared to have an impact on spending; in the higher spending HRRs there were more doctors who did not feel well-prepared to deal with end-of-life care or to communicate with patients about their prognoses and care options and who were more likely to not go to hospice themselves if they were at end-of-life. They were also more likely to recommend chemotherapy for patients with advanced cancers and poor prognoses. In contrast, patient beliefs and preferences had minimal association with spending differences. From a demographic perspective, spending tended to be higher for younger patients, lower for Hispanics, higher for those who had seen a surgeon within six months of diagnosis and higher in areas with lower levels of health resources like hospices. To the extent that physician beliefs and attitudes appear to be a substantial factor in spending variation, helping doctors with more training around end-of-life care would appear to be helpful in limiting variation and ensuring that patient preferences are understood and respected.