Abstract
Dear Editor:
Evaluations of palliative care frequently use end-of-life health care utilization as a key outcome. 1 Hospitals that are early adopters of palliative care programs may differ systematically from U.S. hospitals overall in factors such as hospital size, ownership, and baseline utilization, that may affect programs' effectiveness.
The objective of this study was to describe key characteristics of U.S. hospitals with early palliative care programs compared with those without, including differences in end-of-life utilization.
Methods
We conducted a retrospective cohort study of U.S. hospitals and Medicare admissions. Using American Hospital Association (AHA) hospital characteristics data and Medicare data for beneficiaries who died from 2004 to 2006, we evaluated end-of-life health care utilization in the last six months and one month of life, and linked patients to the last hospital where they were admitted. 2 We included acute care hospitals that responded to the AHA survey with data on palliative care programs, as previously validated. 3 We included a 5% sample of patients with fee-for-service Medicare throughout the study period admitted to a hospital with AHA data in the last six months of life.
We evaluated three key hospital characteristics and seven utilization measures: number of inpatient admissions and hospital length of stay (LOS), number of intensive care unit (ICU) admissions and ICU LOS, and receipt of intubation, feeding tube placement, and CPR. We used chi-squared and t-tests (with log transformation where needed) to compare hospital characteristics and utilization in hospitals with and without palliative care programs.
Results
From 2004 to 2006, the proportion of hospitals reporting a palliative care program increased from 24.3% to 28.4%, with most growth in larger hospitals. Palliative care programs were least common in for-profit hospitals. Among hospitals with <100 beds, 229 (10.8%) reported palliative programs in 2004, compared to 362 (47.6%) of hospitals with >400 beds (p<0.001) (see Table 1).
Utilization rates and significance were almost exactly the same in 2005 and 2006, except rates of in-hospital death were significantly different in both years, e.g., 2006, hospitals with PC programs, 57.1% versus without PC programs, 42.9%; p=0.005.
Results were similar for the one-month analyses in 2004 (utilization less but direction and statistical significance the same) for patients admitted in the last month of life (n=19,606 in 2004), except nonsignificant for feeding tube.
Patients with last admissions to hospitals with palliative programs had significantly more inpatient admissions (2.7 versus 2.5), days in hospital (28.8 versus 25.6), ICU admissions (0.63 versus 0.55), days in ICU (3.9 versus 3.2) and intubations (15.1% versus 11.9%) in the last six months of life than those without palliative programs (p<0.001 for all in 2004) (see Table 1).
Discussion
Hospitals with early palliative programs had statistically significantly higher utilization across multiple measures than those without programs, for both the last six months and one month of life and consistently from 2004 to 2006. We also found high utilization overall and significant variation, consistent with other studies.2,4 Although we did not account for differences in illness, previous studies found that this does not affect regional variation in end-of-life utilization. 5 Other hospital characteristics, such as teaching status, are highly correlated with those we evaluated.
Contextual factors, including baseline utilization, hospital ownership, and bed size, are different in early adopters of palliative care. These factors should be reported in publications of palliative care interventions, as they might impact effectiveness. Future research should evaluate associations of other patient factors and trends over time with utilization in hospitals with palliative care programs.