Abstract
Objective
Determine if hospice enrollment at the time of a terminal admission alters the length of stay (LOS) or costs compared with patients not enrolled in hospice.
Methods
Retrospective chart review of all non-traumatic inpatient deaths of patients with a previous admission in the preceding 12 months at an academic hospital.
Results
209 patients had a non-traumatic death and an admission in the year prior to the terminal admission. Patients enrolled in hospice had a shorter LOS (p = 0.02) and lower cost (p<0.0001) than patients not enrolled at the time of their terminal admission.
Conclusions
Enrollment in hospice during a terminal admission decreased cost and length of stay. Hospice may be a way to provide more cost effective, appropriate care to dying patients.
Keywords: hospice, terminal admission, cost, length of stay, cancer, palliative care consult
Introduction
There is increasing evidence that enrollment in hospice decreases cost, improves patient and family satisfaction, reduces hospital days, and improves the odds that a patient will die in a non-hospital setting.1,2 While patients enrolled in hospice may be more likely to die at home than patients not enrolled in hospice,3 some still die in the hospital for a variety of reasons: including admission for the management of symptoms (e.g. pain crisis or terminal delirium), patient and family preference, and as an unexpected consequence (e.g. someone panics and calls 911).4 We hypothesized that patients who were admitted and died in the hospital while under hospice care had a decreased cost and length of stay compared to people who died in the hospital and did not have hospice support.
Methods
Study Selection and Population
Approval of this study was granted by the University of Iowa Hospitals and Clinics (UIHC) Institutional Review Board. We conducted a retrospective review of all adult inpatient deaths at the University of Iowa Hospitals and Clinics (UIHC) in 2009. Inclusion criteria was an inpatient admission in the previous 12 months. Exclusion criteria was a death due to trauma. For each eligible patient, we recorded age, sex, race, date of death, primary and secondary diagnoses, hospice enrollment, length of stay and total hospital costs on both the final discharge summary and the discharge summary prior to the terminal admission. We also determined which patients were eligible for hospice based on NHPCO criteria at the time of their admission prior to their terminal admission (results in previous publication).5
Statistical analysis
Data were entered in a database using coded identifiers. Statistical analysis was conducted using SAS 9.2 for Windows (SAS 9.2, SAS Institute, Inc., Cary, NC). Standard descriptive statistics were used with wilcoxon rank-sum tests to describe differences in age, cost, and length of stay (LOS) for patients enrolled in hospice versus patients not enrolled in hospice. The LOS and cost of hospice patients were also compared to those of non-hospice patients with diagnoses (primary or secondary) of cancer. This was done with wilcoxon rank-sum tests as well. Wilcoxon rank-sum tests were used because of the small number of hospice enrolled patients (n = 7). Due to this small n, visually comparing the distributions of the samples was difficult, so significant results may be influenced by factors other than those with and without hospice enrollment truly having different medians. To compare the cost and LOS of patients with palliative care vs. those without it, two independent sample t-tests with log transformations of the variables of interest were performed (log transformations were used to better approximate normality of the distributions). The pearson chi square test, as well as the fisher exact test, were used to test for significant differences in proportions of sex, race, primary diagnosis, and type of insurance/payer for patients who were hospice eligible versus those who were ineligible. More specifically, the fisher exact test was used to test proportions which did not meet the assumptions of the pearson chi square test due to small cell sizes. These proportions included terminal diagnosis, race, and type of insurance/payer. The number of days between penultimate and terminal admission were compared between those who were hospice eligible and those who were ineligible as well, and a two independent sample t-test with a log transformation of the variable of interest was used. Spearman correlation was used to determine and test the monotonic relationship between cost observed and the number of secondary diagnoses. Spearman was used in place of pearson correlation due to concerns regarding the normality of the data. Corresponding p-values were recorded, with significance being considered at the standard level of 0.05.
Results
Patient characteristics
This study involved patients who were admitted and died at a large academic institution. The overall demographics and findings are characterized in a previous publication.5 Briefly, of the 688 adult patients who died during 2009, 209 (30%) had a non-traumatic death and a hospital admission in the 12 months preceding the terminal admission. 60% (n = 126) were hospice eligible during the admission proceeding the terminal admission. Days between the previous admission and terminal admission was greater for patients not eligible for hospice (p-value <.0001), with the mean number of days being 62.80 (median 42) for hospice eligible and 130.16 (median 96) for hospice ineligible. There was no significant association between gender, race, or insurance and hospice eligibility status; however, primary diagnosis was associated with it (p-value < .001).
The majority of patients had multiple secondary diagnoses listed on the discharge summary from the terminal admission (number 0 to 8, mean 2.88, SD 1.73). The breakdown of the primary terminal diagnosis is 17% cardiac/vascular (n = 36), 20% respiratory (n = 41), 8% gastrointestinal (n = 17), 13% cancer (n = 28), 3% renal (n = 7), 27% infection (n = 56), 1% rheumatoid/autoimmune (n = 3), and 10% neurological (n = 21).86 of the patients had cancer listed as either a primary or secondary diagnosis.
Patients enrolled in hospice at the time of the terminal admission
Seven patients had enrolled in hospice prior to their terminal admission. All had cancer. Terminal diagnosis listed in the discharge summary included cancer (n=3), respiratory causes (n=3) and sepsis (n=1).
Hospital costs and length of stay
No significant difference in age (p-value = .05) was found between the patients enrolled in hospice versus those not enrolled in hospice. The 7 patients enrolled in hospice demonstrated a decreased cost (p-value < 0.0001) and length of stay (p-value = .02) during the terminal admission compared to patients not enrolled in hospice (Table 1). The same 7 patients did not demonstrate a decreased cost or length of stay during the admission prior to the terminal admission. It is worth noting that no patients were enrolled in hospice during the admission prior to the terminal admission. The decrease in cost and length of stay was maintained when the 7 patients enrolled in hospice were compared with only the other patients that had a diagnosis (primary or secondary) of cancer (p-value < .001, p-value = .0307 respectively). We also identified a correlation between the number of secondary diagnoses and the observed cost. Increased number of secondary diagnosis were correlated with an increased cost (Spearman Correlation value =0.32, p-value <0.001).
Table 1. Comparison of cast and length of stay between patients enrolled and not enrolled in hospice during a terminal hospital admission.
| Enrolled in hospice (n=7) | Not enrolled in hospice, all diagnoses (n=202) | Not enrolled in hospice, cancer diagnosis only (n=79) | |
|---|---|---|---|
| Cost | |||
| Mean | $4,963 | $52,219 | $33,047 |
| Median | $3,690 | $23,322 | $16,091 |
| Standard Deviation | $3,250 | $85,101 | $52,507 |
| Length of Stay | |||
| Mean | 3.43 | 13.35 | 10.51 |
| Median | 1 | 6 | 6 |
| Standard Deviation | 4.47 | 25.05 | 17.49 |
Patients who received a palliative care consult during the terminal admission, also demonstrated a decreased cost (mean cost $41, 859, n=98) but not a decreased length of stay (mean LOS 13.16 days, n = 98) compared to patients who did not receive a palliative care consult (mean cost $58,386, mean LOS 12.89 days, n= 111), with p-values of .04 and .4899 respectively.
Conclusions
Patients enrolled in hospice at the time of a terminal admission demonstrated a striking difference in length of stay and cost when compared to patients who were not enrolled in hospice. This savings was maintained when the pool was decreased to only include cancer patients. This finding is different from Tan's finding that there was no difference in cost when cancer patients are admitted from hospice but his study was not looking at terminal admissions.6 It is similar to previous studies which have shown hospice enrollment demonstrates decreased costs at the end of life as well as decreased length of stay and intensive care utilization while subsequently increasing the quality and satisfaction of the care received.7-12 This may be due to any number of factors, including, the tendencies for hospice patients to have had; goals of care discussions leading to clearer goals of care, a discussion of prognosis leading to an understanding of the terminal nature of the condition, and improved patient monitoring and advocacy. In addition, in support of recent findings regarding palliative care,13 we also confirmed that a palliative care consult, during a terminal admission, has a positive benefit on cost.
This study was exploratory in nature and as such has clear limitations. These include the retrospective study design which was designed to capture only patients who have had a previous hospitalization. It does not include the 479 patients who died in the hospital in 2009 without having a previous admission. The full sample may have included more hospice patients but we feel that is unlikely since repeat hospitalizations can be a marker of disease severity and should make recognition that a patient may be hospice eligible more likely. We also had a very small sample size of hospice patients which may make the results less reliable; however given the robust difference found we think that is unlikely. It would have been interesting to know why patients were being enrolled in hospice but that information was not consistently available in the electronic medical record.
It was striking to us how few inpatients died under hospice care. The majority of the patients had complex medical conditions yet were not enrolled in a beneficial and potentially life extending service. The number of palliative care consults increased from the admission prior to the terminal admission indicating that the primary team was more likely to recognize how ill a patient was during the terminal admission. Unfortunately by this time, it is typically too late for an effective hospice referral. Our previous paper illustrated that hospice eligibility was often not considered even in patients with multiple serious medical illnesses. It is difficult to know how to overcome that hurdle. Educational efforts and further research needs to be done.
This data has implications for the healthcare system. As we strive to provide cost effective, evidence based care, hospice may be an effective tool to meet those objectives in patients with a serious life-limiting illness. Increasing hospice referrals and enrollment rates in appropriate patients is likely to be beneficial to our patients and their families as well as our entire healthcare system.
Acknowledgments
Support for this research was provided by the University of Iowa, Carver College of Medicine, Medical Student Summer Research Fellowship Program (NIH 2-T35-HL007485-31).
Footnotes
Prior Presentations: None
References
- 1.Lim T, Nam SH, Kim MS, Yoon KS, Kim BS. Comparison of Medical Expenditure According to Types of Hospice Care in Patients With Terminal Cancer. The American journal of hospice & palliative care. 2012 May 3; doi: 10.1177/1049909112442586. [DOI] [PubMed] [Google Scholar]
- 2.Emanuel EJ, Ash A, Yu W, et al. Managed care, hospice use, site of death, and medical expenditures in the last year of life. Archives of internal medicine. 2002 Aug 12-26;162(15):1722–1728. doi: 10.1001/archinte.162.15.1722. [DOI] [PubMed] [Google Scholar]
- 3.Taylor EJ, Ensor B, Stanley J. Place of death related to demographic factors for hospice patients in Wellington, Aotearoa New Zealand. Palliative medicine. 2012 Jun;26(4):342–349. doi: 10.1177/0269216311412229. [DOI] [PubMed] [Google Scholar]
- 4.Jeurkar N, Farrington S, Craig TR, et al. Which Hospice Patients With Cancer Are Able to Die in the Setting of Their Choice? Results of a Retrospective Cohort Study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2012 Jun 25; doi: 10.1200/JCO.2011.41.5711. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Freund K, Weckmann MT, Casarett DJ, Swanson K, Brooks MK, Broderick A. Hospice eligibility in patients who died in a tertiary care center. Journal of hospital medicine : an official publication of the Society of Hospital Medicine. 2012 Mar;7(3):218–223. doi: 10.1002/jhm.975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Tan TS, Jatoi A. End-of-life hospital costs in cancer patients: do advance directives or routes of hospital admission make a difference? Oncology. 2011;80(1-2):118–122. doi: 10.1159/000328279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Carlson MD, Herrin J, Du Q, et al. Impact of hospice disenrollment on health care use and medicare expenditures for patients with cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2010 Oct 1;28(28):4371–4375. doi: 10.1200/JCO.2009.26.1818. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Taylor DH, Jr, Ostermann J, Van Houtven CH, Tulsky JA, Steinhauser K. What length of hospice use maximizes reduction in medical expenditures near death in the US Medicare program? Soc Sci Med. 2007 Oct;65(7):1466–1478. doi: 10.1016/j.socscimed.2007.05.028. [DOI] [PubMed] [Google Scholar]
- 9.Zhang B, Wright AA, Huskamp HA, et al. Health care costs in the last week of life: associations with end-of-life conversations. Archives of internal medicine. 2009 Mar 9;169(5):480–488. doi: 10.1001/archinternmed.2008.587. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Pyenson B, Connor S, Fitch K, Kinzbrunner B. Medicare cost in matched hospice and non-hospice cohorts. Journal of pain and symptom management. 2004 Sep;28(3):200–210. doi: 10.1016/j.jpainsymman.2004.05.003. [DOI] [PubMed] [Google Scholar]
- 11.Norton SA, Hogan LA, Holloway RG, Temkin-Greener H, Buckley MJ, Quill TE. Proactive palliative care in the medical intensive care unit: effects on length of stay for selected high-risk patients. Crit Care Med. 2007 Jun;35(6):1530–1535. doi: 10.1097/01.CCM.0000266533.06543.0C. [DOI] [PubMed] [Google Scholar]
- 12.Abernethy AP, Kassner CT, Whitten E, Bull J, Taylor DH., Jr Death service ratio: a measure of hospice utilization and cost impact. Journal of pain and symptom management. 2011 Jun;41(6):e5–6. doi: 10.1016/j.jpainsymman.2011.03.004. [DOI] [PubMed] [Google Scholar]
- 13.Morrison RS, Dietrich J, Ladwig S, et al. Palliative care consultation teams cut hospital costs for Medicaid beneficiaries. Health affairs. 2011 Mar;30(3):454–463. doi: 10.1377/hlthaff.2010.0929. [DOI] [PubMed] [Google Scholar]