|Home | About | Journals | Submit | Contact Us | Français|
To evaluate factors that affect the financial performance of hospice.
Using the California Office of Statewide Health Planning and Development 2003 survey, we evaluated the organizational attributes, clinical care, and financial performance of 185 operational hospices. As outcomes, we evaluated revenues, costs, and profits per patient and per patient–day, the intensity and skill mix of care, and the provision of charitable and special palliative services. We evaluated regression-adjusted differences by profit status controlling for other organizational features and aggregate patient characteristics.
Hospices reported median revenue of $6865 per patient and $138 per patient–day (for-profit-not-for profit [FP-NFP] difference −$20, p = 0.045), median cost of $6737 per patient, and $135 per patient–day (FP-NFP difference −$55, p = 0.002), and median pretax profit of $334 per patient and $6 per patient–day (FP-NFP difference $34, p = 0.026). Patients received a median of 29.9 total visits by all providers per patient (FP-NFP difference 8.8 visits, p = 0.010), but there was no difference in total visits per patient–day. A median of 50.8% of all nursing visits were registered nurse (RN) visits (FP-NFP difference −14.1%, p<0.001). Few hospices provided charity care, and only 4% of hospices reported expenditures on chemotherapy and only 9% on radiation therapy.
Overall hospice profitability is low. Length of stay is strongly associated with financial performance, and greater FP profitability is related to lower costs. FP hospices also provide less RN care as a proportion of nursing care. Few hospices provide charitable care or special costly services. The relationship of service patterns to patient quality needs to be examined.
Americans share a broad consensus about factors important to achieving better end of life care, and too few Americans achieve these ends in dying.1 A major concern is that our health care system makes few accommodations to help patients and families address their goals, which include symptom management, ensuring spiritual well-being, supporting caregivers, and advance care planning. The last year of life consumes almost 30% of lifetime Medicare expenditures, and yet many patients do not receive adequate care during that time.2 Hospice is the primary means available to help Americans achieve better late life care, and recent evidence supports the effectiveness of hospice and similar services.3,4
Important limitations have been described in access to hospice, and there have been recent calls to address them through financing and regulation.5 Even among patients who receive hospice, enrollment is typically late, with a median stay of 26 days in 2005.6 Palliative needs are substantial in certain settings-such as nursing homes-and in conditions other than cancer that are relatively underserved by hospice.5 To understand the implications of modifying the benefit, even basic information about the financial performance of hospices and how these factors relate to the sustainability of the current model are lacking.
We previously examined these factors in the context of a comparison of for-profit (FP) and not-for-profit (NFP) hospice providers.7 For-profit hospices grew to 27% of Medicare-certified hospices in 1999, and several large chains have recently gone public.8 We found that the patients of FP providers had more stays over 90 days in length and received more nursing visits, but less RN care, and that these differences were associated with the characteristics (e.g., long-term care referrals, noncancer diagnoses) of providers' patient populations.7 A recent MedPac analysis suggests those differences may be related to hospices trying to achieve longer lengths of stay (LOS), because services are more intense at the beginning and end of stays, and therefore profitability is greater with longer, relatively stable stays;9 however, the relationship of these issues to financial performance has not been examined.
Using the 2003 California Office of Statewide Health Planning and Development (OSHPD) annual survey, we obtained organizational descriptors, aggregate patient data, and financial information (derived from cost reports).10 We examined factors associated with the financial performance of FP-NFP providers including LOS. To further illuminate FP-NFP differences, we evaluated the association of profit status to the intensity and skill mix of nursing care, because greater use of skilled care has been associated with higher quality of care in other settings.11 We also examined associations of profit status with charitable care and the provision of special or high-cost palliative services.
California OSHPD annually conducts a statewide survey of licensed hospices. Completed individual surveys for 2003 covering the report period January 1 to December 31 are available in hypertext markup language (HTML) format on the OSHPD website,10 so we used custom Practical Extraction and Report Language (PERL) scripts to build a basic analytic dataset. Although the responses on the public use site were largely complete, in certain cases (<5%) we verified or corrected erroneous, missing, or ambiguous survey items via telephone follow-up.
We analyzed total revenue, total revenue minus grants and donations, total cost, pretax profit, and pretax profit minus grants and donations. Total revenue is revenue from all sources, including clinical revenue, grants, donations, and unrelated business income, minus contractual adjustments from payers, denials and bad debt, and tax write-offs (including charity care). Cost is the total reported organizational cost. Pretax profit is total revenue minus total cost. By examining revenue and profit with and without charitable grants and donations, we were able to evaluate the extent to which they contribute to hospices' financial performance. In order to examine the association of LOS with financial performance, we evaluated all outcomes on both a per-patient and per-patient–day basis.
We evaluated average LOS, total visits by all types of providers, the proportion of nursing visits conducted by RNs, and the proportion of total provider visits conducted by physicians. For discharged patients, hospices reported total patients and total patient–days, so following our previous methods,7,12 we calculated average LOS as total patient–days for discharged patients divided by the total number of discharged patients. We defined total provider visits as the sum of patient visits made by RNs, licensed vocational nurses (LVNs), social workers, physicians, homemakers and home health aides (HHAs), chaplains, and other clinical staff. As a measure of care intensity, we calculated total provider visits both per patient and per patient–day. We defined total nursing visits as the sum of RN, LVN, homemaker and HHA visits, and calculated the skill mix of nursing visits as RN visits divided by total nursing visits. We also evaluated physician visits as a proportion of total provider visits.
In return for their tax-advantaged status, NFPs may return resources to the community through education, charity care, higher quality care, or other valued services. Due to the extremely skewed distribution of hospices' reported expenditures for charitable care, we categorized hospices as having provided any charity care versus none. Special palliative services included bereavement, pediatric programs, adult day care, and palliative inpatient units. We summed the responses and categorized each hospice as making available 0, 1, or 2 or more special services. We evaluated whether or not hospices reported having any costs associated with delivering chemotherapy or radiotherapy. Both treatments can play an important palliative role in selected cases, yet they are costly and may not always be available in hospice.12,13
We categorized “investor”-type hospices (corporation, limited liability company, partnership, or individual) as FP, and all others (city/county, district, nonprofit corporation (including church related), University of California, or state) as NFP. Other organizational characteristics included size, time in operation, urbanicity, and corporate integration. Size was defined as total unduplicated patients in 2003. Time in operation was calculated as a continuous variable in years from the hospice's initial license date until December 31, 2003. Hospices reported their service areas as urban, rural, or mixed urban and rural. Hospices reported their agency type (freestanding, home health-based, hospital-based, long-term care facility-based, or Veterans Administration-based) and entity relationship (sole facility, branch, or parent with branches), and we combined these to define four categories of integration: freestanding, sole facility; freestanding, branch or parent; home health care agency-based; and hospital-based. We also controlled for aggregate patient characteristics, including referral source, insurance coverage, percentage of noncancer diagnoses, and percentage of non-white patients. With the exception of size (in this study we used total patients instead of total patient–days), these definitions were consistent with definitions used in our previous analyses.7,12,14,15
We performed bivariate analyses using either Wilcoxon (for continuous variables) or χ2 (for categorical variables) tests of equality.16 Bivariate analyses examined differences in organizational and aggregate patient characteristics by (1) profit status, (2) the availability of financial data in the OSHPD report, and (3) whether or not hospices began operating after January 1, 2003. We analyzed regression models that controlled for profit status as well as other organizational and aggregate patient characteristics. To evaluate the relationship between charitable inputs and care, models of the clinical performance and charitable output measures also controlled for financial contributions and grants received as well as donated volunteer hours.
We calculated regression-adjusted differences by profit status using linear regression for all outcomes, except for whether a hospice provided any charity care and for the number of special services provided. For these outcomes, logit and generalized ordered logit models were estimated, yielding relative risks associated with FP status (the ratio of the adjusted probabilities of the outcome for FPs versus NFPs). Square root transformations of LOS were used due to skewness, so we used a “smearing” retransformation algorithm to calculate the predictive margins associated with the independent variables. Smearing adjusts the predictions so that the mean of the predicted (regression-adjusted) values equals the mean of the actual (unadjusted) values.17 Confidence intervals around the relative risks and predictive margins were calculated using bias-corrected empirical bootstrapping with 1000 replicate samples. All statistical tests used a 5% significance level except where noted. Heteroskedasticity was indicated by White, Breusch-Pagan, and Cook and Weisberg's tests, so we computed all linear regressions with robust standard errors. For the multivariate models, we singly imputed 10 missing values for the percentage of non-white patients. We used SAS 9.1 (SAS Institute, Cary, NC) and STATA/SE 8.2 (StataCorp, College Station, TX) for all analyses.
One hundred eighty-five agencies served at least one patient in 2003. Because they do not report financial data, 13 Kaiser affiliates were ineligible, and of the remaining 172 hospices, 13 failed to report 2003 financial data, and 1 did not provide patient data. Bivariate analyses revealed no significant differences among hospices with complete data versus those excluded.
Differences by profit status in the median values for organizational and patient characteristics are shown in Table 1. Compared to NFP hospices, FPs served fewer patients (240 versus 332, p = 0.04), opened more recently (4.9 versus 8.3 years, p<0.001), had a longer average LOS (57.9 versus 46.4 days, p<0.001), received less charitable revenue ($0 versus $5988, p<0.001), used less volunteer labor (844 versus 1949 hours, p<0.001), were more likely to serve a primarily urban area (75% versus 56%, p = 0.02 for group), less likely to serve a primarily rural or mixed rural/urban area (7% versus 13% rural, 27% versus 31% mixed, p = 0.02 for group), more likely to be a freestanding, sole agency (56% versus 26%, p<0.001 for group) and less likely to be hospital or home-health based (2% versus 26% for hospital, 19% versus 25% for home-health; p<0.001 for group). FP agencies also cared for more Medicare-insured patients (92.7% versus 89.3%, p = 0.02), referrals from long-term care (15.7% versus 5.9%, p<0.001), non-white patients (21.3% versus 13.5%, p = 0.02), and discharged patients with non-cancer diagnoses (60.1% versus 52.1%, p = 0.01).
Median values for the financial performance measures and their regression-adjusted associations with profit status are shown in Table 2. Including grants and donations, median revenue was $6865 per patient, and $138 per patient–day. Subtracting grants and donations, median revenue was $6452 per patient, and $132 per patient–day. Median cost was $6737 per patient, and $135 per patient–day. Median pretax profit was $334 per patient and $6 per patient–day. Without grants and donations, median profit was $63 per patient and $1 per patient–day.
After adjusting for observable differences, FPs did not have significantly different revenues per patient than NFPs (+$456, p = 0.38), but received $20 less per patient–day (p = 0.05). After subtracting grants and donations from overall revenues, FPs earned $1076 more per patient (p = 0.03) and had similar revenues per patient day (+ $6, p = 0.53). FPs reported $55 less in costs per patient–day (p = 0.002). FPs reported $1473 more pretax profit per patient (p = 0.03) and $34 more pretax profit per patient–day (p = 0.03). Without taking into account grants and donations, FPs would net $2103 more per patient in pretax profit (p = 0.003) and $49 more per patient–day (p = 0.001).
Median values of the clinical performance measures and their regression-adjusted associations with profit status are shown in Table 3. Average LOS was 12.0 days longer for FPs than NFPs (p = 0.002). Hospices made a median of 29.9 total provider visits per patient; compared with NFPs, FPs reported 8.8 more total provider visits per patient after regression adjustment (p = 0.01), but there was no difference in total provider visits per patient–day (p = 0.94). The median proportion of RN visits as a proportion of all nursing visits was 50.8%; compared to NFPs, FPs reported 14.1 percentage points fewer RN visits (p<0.001). As a sensitivity analysis, we found a similar association of profit status with total RN visits. Hospices reported so few physician visits in 2003 (1.4% of all visits overall; 1.6% FP, 0.8% NFP, p<0.001) that multivariate comparisons by profit status were not possible.
In results not shown in the tables, there were no significant differences between FP and NFP hospices in the provision of charity care (relative risk [RR]: 0.69, 95% confidence interval [CI]: 0.30–1.94). FPs were 58% more likely to indicate the availability of only 1 special service (RR: 1.58, 95% CI: 1.01–28.05). Although not statistically significant, FPs were more likely to indicate the availability of no special services (RR: 1.21, 95% CI: 0.53–3.12) and less likely to indicate the availability of 2 or more special services (RR: 0.30, 95% CI: 0.01–1.19). Overall, only 4% and 9% of hospices reported expenditures on chemotherapy and radiation therapy, respectively. Although multivariate comparisons by profit status of these outcomes were not possible, bivariate results were suggestive of differences (for chemotherapy use, FP = 0.0%, NFP = 5.7%, p = 0.08; for radiation use, FP = 1.9%, NFP = 12.3%, p = 0.03). NFPs accounted for nearly all of the spending on these treatments in 2003, as 6 NFPs spent a total of $43,590 (average $7265) on chemotherapy and 13 NFPs spent a total of $154,647 (average $11,896) on radiation treatments.
In summary, overall hospice industry profitability is low, with a median annual organizational profit of $75,683 (range: −$2,444,719 to $2,889,763). However, FP hospices are more profitable than NFPs. Their financial advantage appears to be driven by longer LOS. While FPs provided more visits per patient, this was also due to longer LOS, not a greater intensity of care. In contrast, FPs provided a lower proportion of RN visits to total nursing visits than NFPs. FP profitability is reflected in taxes paid—and among those FP hospices that were profitable, 20.5% of pretax profits were paid in taxes. There were no differences between FP and NFP hospices in the provision of charitable care, and little difference in the availability of special services. Regardless of profit status, few hospices provided physician visits, chemotherapy, or radiotherapy.
Limitations of our analysis include the fact that we only examined California, because the detailed organizational data included in the OSHPD survey are not readily available elsewhere. However, hospice policy is dominated by Medicare, and the hospices we studied served a large number of all 2003 U.S. hospice patients (66,884 of 950,000 total estimated patients) and thus are important in their own right.6 We examined financial data on a per-patient and per-patient–day basis to account for potential biases from including 11 recently opened and 3 recently closed hospices. However, we found that recently opened hospices were more likely to be FP, with lower revenues and higher costs than others, suggesting a conservative bias. Finally, patient-level information about the quality of care is needed to evaluate the implications of our findings.
We found that longer LOS was strongly associated with differences in overall profitability, and this factor was associated with caring for patients with noncancer diagnoses and those from nursing homes. In 2003, 50.6% (33,818 of 66,884) of California hospice patients had noncancer primary diagnoses, an increase from just 35% in 1997.7 Nationwide, 41.8% of hospice patients had noncancer primary diagnoses in 2000.18 This trend may imply that hospice access is improving among patients with diagnoses, other than cancer. Non-cancer patients have less predictable short-term prognoses.19 Adjusting for the other covariates, the average LOS of a hospice would increase by 7.11 days (p = 0.004) if the percentage of non-cancer diagnoses were increased from the 25th percentile value (42.9%) to the 75th percentile value (64.1%).
Thus, individual hospices may benefit from enrolling patients with non-cancer diagnoses, and patients may also benefit—hospice penetration of nursing homes overall is low and the palliative needs of US nursing home patients are extensive.20–22 Our current study adds to the evidence of prior research that nursing homes are an important source of hospices' referrals. The proportion of noncancer patients was highly correlated with the proportion of hospice patients referred through long-term care (Pearson r = 0.42), but in multivariate analysis, only the proportion of non-cancer diagnoses was independently associated with LOS.
Because profit status is also associated with differences in clinical performance including the duration of care (e.g., LOS), frequency of visits, and how often RNs compared to other nursing providers (e.g., LVNs, homemakers) are used, further research is needed to assess differences in quality outcomes. We found here as in our previous work that differences in hospice patient population partially account for differences in staff utilization among hospices. Supporters of FP health care argue that services are delivered more efficiently, while opponents suggests that this financial advantage comes at the cost of lower quality care.24–26 The question of what the “right” LOS, nursing intensity, and provider skill mix is for hospice care remains unclear and answering this question will require research that examines variation in these factors with patient clinical outcomes. Furthermore, the nursing shortage could play a role in how hospices are using their staff. Future research should consider the substitution effect more broadly and account for local nursing supply and wages.
NFP hospices have lower pretax profit than FPs, but that difference would be even greater without charitable grants and donations. Because hospice is a low-profit industry overall, charitable income is important to the financial viability of NFP hospices. However, the amount of charity care delivered did not differ by profit status. Charitable revenue is expected to allow NFPs the financial freedom to provide unprofitable services for the community, but NFPs are generally unprofitable entities even with charity revenue. The amount of charitable contributions supporting NFP hospices as a whole could be seen as adequate for financial survival, but insufficient for true fulfillment of the NFP mission.
The low profitability of the hospice industry suggests that hospices will continue to be challenged to provide newer, even more costly palliative therapies. That suggests that hospice patients may not have adequate access to important palliative modalities—or that patients will only enter hospice after they and their providers have exhausted such treatments, keeping LOS extremely short. Payment for other services provided for the terminal illness has to come from the fixed per diem Medicare payment. Hospices infrequently provide chemotherapy and radiation, but this is understandable given the median profit of $6 per day including grants and donations, Addressing this issue will be necessary if it is desirable to integrate hospice more “upstream” in the chronology of illness.
In conclusion, our report provides important information that has been lacking about the financial performance of hospice and sheds light on how policy might influence factors that affect the profitability and financial viability of hospices. As supported by other research, we found that LOS is a particularly important factor associated with differences in financial performance.9 Because of looming demographic changes, cost-effective, high-quality options are urgently needed to assure adequate access to palliative care across the continuum of serious illness care.
Mr. O'Neill's work was supported by the NIA/AFAR and Lillian R. Gleitsman Summer Training Institute at UCLA. Dr. Lorenz is a VA HSR&D Advanced Career Development Awardee.