Profit Status of Home Health Care Agencies and the Risk of Hospitalization

Abstract

We examined the association of profit status and patient hospitalizations in the present-day home health care market, a market that grew substantially in the past decade, with much of that growth attributed to the entry of for-profit agencies. Data from the 2007 National Home and Hospice Care Survey were linked to the risk-adjusted agency-level measure of the percent of home health episodes of care ending in hospitalizations available from the Centers for Medicare and Medicaid Services' (CMS) Home Health Compare Web site. A linear regression model was estimated (n = 510). Control variables included other agency characteristics besides profit status, area hospital bed supply, and state dummy variables to control for state fixed effects. For-profit agencies were more likely than not-for-profit agencies to have a risk of hospitalizations greater than expected after accounting for patient characteristics and model control variables. Attributes of the CMS hospitalization measure are discussed and implications for future research described. (Population Health Management 2011;14:199–204)

Background

F ew studies have investigated the comparative performance of for-profit and not-for-profit home health care agencies (HHAs),¹ which is in stark contrast to the volume of research comparing for-profit and not-for-profit nursing homes.² Hospitalizations of nursing home residents, in particular, has been studied because hospitalizations are costly and may cause harm as a result of disruptions in care and hospital-acquired illnesses.^3,4 Analysts have argued that for-profit nursing facilities are more likely than not-for-profit nursing facilities to hospitalize residents to reduce expenditures from treating sick residents.⁵ Findings show for-profit nursing homes are more likely to hospitalize residents than not-for-profit facilities.⁴ Similar financial incentives to hospitalize patients might exist among for-profit HHAs. What evidence exists on HHAs, however, indicates no association between profit status and hospitalizations.^6,7 In this study, we provide additional analysis of the relationship between HHA profit status and hospitalizations.

Framework

In recently published studies, neither Grabowski et al⁶ nor Smith et al⁷ found profit status associated with the hospitalizations of home health (HH) patients.^6,7 Smith et al⁷ studied hospitalizations specifically, whereas Grabowski et al⁶ studied the outcome of discharges to institutional settings, which included hospitals. The most recent time period covered by both studies was the year 2000 using data on discharged HH patients from the 2000 National Home and Hospice Care Survey (NHHCS).^6,7

Wide fluctuation in the number of HHAs was associated with the implementation of the Balanced Budget Act of 1997, which altered the Medicare payment system for HHAs. Initial implementation of the Act's provisions resulted in an approximate 30% drop in the number of Medicare-certified HHAs by the early 2000s from the 10,000 in 1996, with much of the decline among for-profit agencies.^8,9 Whether this fluctuation affected the characteristics of for-profit agencies in earlier studies using 2000 data^6,7 is unknown, but it is a possibility. Associated with modifications to the payment system, the number of HHAs grew back to over 9000 agencies in 2007 with most of the growth attributed to the entry of for-profit agencies.^9,10

In this study, the question of the association between HHA profit status and hospitalizations is revisited using data from the 2007 NHHCS,¹¹ data from a period following the recent growth in HHAs. The measure of the risk of hospitalization comes from the Centers for Medicare and Medicaid Services' (CMS) Home Health Compare Web site.¹² As will be discussed, a favorable property of this measure of hospitalizations is its inclusion of not only discharges to hospitals but also occurrences in which HH patients were transferred for a hospital stay but not discharged from the HHA.

Methods

Data sources

Three data sources were used. (1) Data on HHA characteristics came from the 2007 NHHCS.¹¹ (2) Data on hospitalizations from HHAs were downloaded from CMS Home Health Compare.¹² (3) A measure of county-level hospital bed supply was derived from the Area Resource File (ARF).¹³

The 2007 NHHCS is a nationally representative sample of home health and hospice care agencies. Each agency in the NHHCS is associated with a sampling weight based on the agency's sampling probability from the universe of home health and hospice care agencies in the United States. Data were collected during interviews conducted at agencies between August 2007 and February 2008. The sample included agencies that provided only HH (HH only), only hospice care, and agencies that provided both home health and hospice care (“mixed agencies”). The 2007 NHHCS is described in detail elsewhere.¹¹

The CMS Home Health Compare data, described in the following paragraphs, is for Medicare-certified HHAs. Of the 677 HHAs in the 2007 NHHCS sample, 624 were Medicare certified. Among the 624 Medicare-certified HHAs, 37 could not be matched to the Home Health Compare data, and another 77 were missing data on 1 or more agency variables in this study, yielding 510 cases in the analysis. The 510 cases represent 81% of the population (weighted total) of Medicare-certified HHAs estimated from the 2007 NHHCS. There were no significant differences in profit status, chain affiliation, or agency type—the three attributes with no missing data—between the Medicare-certified agencies in the study and those with missing data.

Dependent variable

The measure of annual risk of hospitalization from the CMS Home Health Compare—a quality measure endorsed by the National Quality Forum¹⁴ —is an adjusted risk among episodes of care occurring during a moving 12-month period. Based on a 12-month period, the measure incorporates any seasonal variation. The measure is updated quarterly on the Home Health Compare Web site with a new moving 12-month period. In the CMS measure, an episode of care starts at admission or resumption of care and ends with a discharge or with a transfer to an inpatient setting for an overnight stay; hence, the measure captures both formal discharges to hospitals and transfers for a hospital stay without being discharged from the HHA.¹⁵ The measure of hospitalizations was developed from the patient assessments in the Outcome and Assessment Information Set (OASIS) completed in Medicare-certified HHAs for Medicare and Medicaid HH patients.^12,16

The measure of hospitalizations adjusts for expected risk with a regression model that predicts a patient's probability of hospitalization; the model includes patient demographics, payer, and clinical comorbidity covariates.¹⁶ The predicted prevalence of hospitalizations at an agency is the sum of predicted probabilities across agency patients (ie, episodes). The adjustment entails subtraction of an agency's predicted prevalence from the observed prevalence. To avoid negative numbers from this subtraction and to standardize the comparison across agencies, an agency's adjusted prevalence is added to a constant value (in this case, the national predicted HHA hospitalization rate).¹⁶ Agencies with more observed hospitalizations than predicted are above the constant and those with fewer hospitalizations than predicted are below the constant. Hence, the scale in the measure represents the comparative risk or likelihood of hospitalization after adjusting for expected risk due to agency case-mix.

For each agency, the 12-month period used for the measure of hospitalizations started approximately at the time of the NHHCS data collection. For example, if an agency's data were collected during October, November, or December 2007, the 12-month prevalence of hospitalizations used from Home Health Compare started within the 4^th quarter of 2007.

Previous studies on hospitalizations by HHAs only included hospitalizations at discharge.^6,7 Our analysis of the 2000 NHHCS discharge sample of Medicare and Medicaid HH patients discharged alive from Medicare-certified HHAs showed, for example, that the percent of discharges to hospitals was 19% among for-profit agencies and 16% among not-for-profit agencies. The 2007 NHHCS included a sample of current HH patients at participating HHAs at the time of the NHHCS agency interview but did not include a discharge cohort. For the first time in the NHHCS series, the 2007 NHHCS asked agency staff whether sampled patients had a hospital stay since admission to the HHA without being formally discharged from the HHA. The percent of current patients reported to have had transfers to hospitals in for-profit and not-for-profit Medicare-certified HHAs were 18% and 24%, respectively. Clearly, neither the discharge status of a discharge cohort nor the transfers of a current patient cohort alone appears to sufficiently cover the occurrences of hospitalizations of HH patients within either for-profit or not-for-profit agencies. As noted, the CMS measure of hospitalizations captures hospitalizations associated with both discharges and transfers.

Profit status

Profit status has 2 categories: for-profit or not-for-profit. Not-for-profit agencies include private nonprofit agencies and government-operated agencies.

Other agency characteristics

Control variables were included. Number of current patients is a continuous variable. Agency location in a metropolitan statistical area (metropolitan versus non-metropolitan area), agency type (HH only versus mixed agency), and chain affiliation (chain versus non-chain) are binary variables.

Another variable differentiated independently owned/operated agencies from agencies owned/operated by another provider type, such as a hospital or nursing home. This variable is different from chain affiliation. For example, a HHA may be part of a chain of agencies that provides only home health/hospice care and, thus, is not owned/operated by a hospital or nursing home.

A variable indicating the percent of an agency's patient revenue from Medicaid payments was included because dependence on Medicaid payment has been shown to be a factor in the risk of hospitalizations among nursing home residents.^4,5 The 2007 NHHCS did not include a measure of the size of the Medicaid patient census. In addition, state dummy variables were included to control for state fixed effects such as any differences in Medicaid payment and service coverage policies.

The staffing levels of registered nurses (RNs) and certified home health aides (aides) may relate to case-mix and hospitalization risk (eg, a greater number of RNs may indicate a greater proportion of patients with complex clinical needs). Therefore, one variable included is the number of patients per RN full-time equivalent (FTE) position and the other is patients per aide FTE.

The 2007 NHHCS included questions on the telemedicine capacities of an agency, capacities that may affect risk of hospitalizations. A variable was included indicating whether the agency had the ability to use at least routine non-video patient monitoring that did not require conversation (eg, regular transmission of vital signs).

Variables indicating services an agency provided (yes or no) were included as additional risk factors: physical therapy, speech therapy, intravenous (“IV”) therapy, wound care, respiratory therapy, physician services, and homemaker services. Occupational therapy was not included because of its high correlation with physical therapy (r = 0.86).

Hospital bed supply

Hospital use may also be affected by local health care resources.¹⁷ The county-level number of inpatient hospital beds per 1000 persons 65 years of age and older was included using ARF data.

Statistical analyses

Descriptive statistics and the linear regression model on risk of hospitalizations were estimated using SUDAAN¹⁸ developed for analysis of stratified sampling designs.

Results

Descriptive statistics

An estimated 65% of Medicare-certified HHAs are for-profit (Table 1). For-profit agencies are more likely than not-for-profit agencies to be HH-only agencies rather than mixed agencies. The mean prevalence of hospitalizations across all agencies was 31% (Table 1). The mean annual adjusted risk of hospitalizations among for-profit agencies is greater than that of not-for-profit agencies by over 4 percentage points.

Table 1.

Descriptive Statistics

	All Agencies (n = 510)		For-Profit^a (n = 174)	Not-for-Profit (n = 336)
Categorical Variables	Percent^b	(S.E.)	Percent^b	Percent^b
For-profit	64.96	(4.11)	—	—
Chain	30.49	(4.61)	41.79	9.52^***
Home-health-care only	87.01	(1.77)	96.55	69.33^***
Independent owner/operator	70.30	(3.76)	92.27	29.59^***
Non-video patient monitoring	14.85	(2.32)	8.98	25.72^***
Agency Services
Physical therapy	91.50	(2.88)	89.94	94.41
Speech therapy	82.90	(3.86)	82.96	82.78
IV therapy	87.75	(3.08)	88.67	86.05
Wound care	97.28	(1.33)	97.02	97.75
Respiratory therapy	17.99	(3.65)	16.51	20.73
Physician services	10.92	(1.87)	7.01	18.17^**
Homemaker services	27.06	(3.81)	26.52	28.07
Metropolitan area	72.27	(2.59)	79.30	59.24^**

Continuous Variables	Mean^c	(S.E.)	Mean^c	Mean^c
Risk of hospitalizations	31.15	(0.74)	32.76	28.17^**
Percent revenue from Medicaid	12.26	(1.79)	10.65	15.24
Number of patients	142.71	(11.83)	99.96	221.97^***
Patients per RN FTE	15.05	(0.84)	15.58	14.07
Patients per aide FTE	47.61	(4.83)	38.98	63.61^*
Hospital beds per 1,000 65+	3.25	(0.13)	3.28	3.20

IV, intravenous; RN FTE, registered nurse full-time equivalent position; aide FTE, certified home health aide full-time equivalent position, 65+, persons 65 years of age or older in the population.

The NHHCS sample sizes for home-health-care only, hospice-care only, and mixed agencies are similar. Thus, given the universe of home-health-care only agencies is greater than that of hospice-care only and mixed agencies, the sample weights—which are the reciprocal of the sampling probability—are, on average, greater for the home-health-care only cases. A greater percent of home-health-care only agencies are also for-profit, which, with the greater weights, appropriately results in the smaller sample of for-profit agencies representing an estimated 64.96% of the population of Medicare-certified home health care agencies.

weighted percent.

weighted mean.

P < 0.05 for t test of difference between for-profit and not-for-profit.

P < 0.01 for t test of difference between for-profit and not-for-profit.

***

P < 0.001 for t test of difference between for-profit and not-for-profit.

Table 1 shows that RNs have fewer patients per FTE than aides, which is consistent with CMS information showing the total number of RN FTEs in Medicare-certified HHAs nearly double that of aides.⁹ RNs may provide most of the care in these agencies, which differs from nursing homes where nursing assistants provide most of the care.¹⁹ Not-for-profit agencies have a greater mean number of patients per aide FTE.

Regression model

As shown in Table 2, the adjusted risk of hospitalizations was significantly greater (by 4.5 percentage points) in for-profit agencies relative to the risk in not-for-profit agencies. The adjusted risk of hospitalizations in for-profit agencies was greater after accounting for other covariates that may vary between for-profit and not-for-profit agencies and may also relate to risk of hospitalizations.

Table 2.

Regression on Agency Annual Risk of Hospitalizations (n = 510)

	Hospitalizations
Variable^a	Coefficient	(P-value)
For-profit	4.478	(<0.001)
Chain	0.532	(0.666)
Percent revenue from Medicaid	0.026	(0.365)
Number of patients	−0.001	(0.782)
Home-health-care only	1.483	(0.222)
Independent owner/operator	0.214	(0.859)
Patients per RN FTE	0.023	(0.705)
Patients per aide FTE	−0.019	(0.013)
Non-video patient monitoring	−2.305	(0.014)
Agency Services
Physical therapy	5.450	(0.031)
Speech therapy	−5.255	(0.005)
IV therapy	3.525	(0139)
Wound care	−4.723	(0.032)
Respiratory therapy	3.725	(0.002)
Physician services	1.058	(0.411)
Homemaker services	−3.014	(0.022)
Metropolitan area	−1.352	(0.181)
Hospital beds per 1,000 65+	0.137	(0.713)
Intercept	18.795	(<0.001)
R²	0.518^b	(<0.001)
Adjusted R²	0.447^b	(<0.001)

IV, intravenous; RN FTE, registered nurse full-time equivalent position; aide FTE, certified home health aide full-time equivalent position; 65+, persons 65 years of age or older in the population.

A dummy variable was included in the model for each state and the District of Columbia, excluding 1, to control for state fixed effects; coefficients on these variables are not reported in the table.

The state variables controlling for state fixed effects account for a sizable part of the variance explained. In a model without the state variables, R ² = .251; Adjusted R ² = .224.

Other covariates were also significantly related to hospitalizations at P < 0.05. HHAs that provide physical therapy, for example, were associated with a greater risk of hospitalizations relative to agencies that do not provide physical therapy. Similarly, providing respiratory therapy was associated with a greater risk of hospitalizations relative to agencies not offering respiratory therapy. HHAs with homemaker services were associated with a lower risk of hospitalizations relative to HHAs without these services. Agencies that indicated they had telemedicine capacity for routine nonvideo monitoring of patients also had a lower risk of hospitalizations compared to agencies without this capacity. Providing speech therapy and providing wound care were each significantly associated with lower hospitalization risk. Unanticipated was the negative association of number of patients per aide FTE with hospitalizations. Additional research is needed to clarify how the staffing ratio for aides might relate to HHA practices or case-mix affecting risk of hospitalizations.

Discussion

Among HHAs, for-profit status was associated with a greater risk of hospitalizations than expected relative to the risk associated with not-for-profit status. Previous studies found no association between the profit status of HHAs and hospitalizations.^6,7 Using data covering a more recent time period from the 2007 NHHCS,¹¹ coupled with the CMS measure of hospitalizations that captures hospitalizations associated with both discharges and transfers,^12,15 we did find for-profit status of HHAs related to a greater risk of hospitalizations.

A limitation is that the findings are based only on Medicare and Medicaid HH patients in Medicare-certified HHAs. Nonetheless, analysis of the 2007 NHHCS showed that Medicare-certified HHAs in 2007 represented 82% of all HHAs. Also, Medicare and Medicaid HH patients represented an estimated 83% of all HH patients at any time in these facilities. The findings, of course, assume that the case-mix adjusted measure of hospitalizations accounts adequately for relevant case-mix differences between for-profit and not-for-profit agencies.

In a recent study, Schade and Brehm argue that the CMS measure of hospitalization used in this study has an agency-level average length-of-stay bias.²⁰ They argue that for-profit agencies may have a greater risk of hospitalizations because for-profit agencies have a greater average length of stay. Schade and Brehm did not include profit status in their reported stepwise regression results because they noted it was not significant when agency average length of stay was included.²⁰

An agency's average length of stay is a provider characteristic, not a patient characteristic. Length of stay is known after an episode of care, not beforehand. The earlier risk adjustment of the CMS home health hospitalization measure included length of stay as a risk factor.²¹ The present risk adjustment of this measure does not include length of stay; the intent was to include only patient risk factors at admission to a HHA.¹⁶

In their analysis of OASIS data, Schade and Brehm²⁰ used the risk adjustment reported by Shaughnessy and Hittle,²¹ which is the earlier risk adjustment including each patient's length of stay. Hence, the risk of hospitalization predicted by agency-level length of stay in the Schade and Brehm study was the risk that included adjustment for patient-level length of stay.²⁰ As Schade and Brehm mention, length of stay of the home health episode in the earlier patient-level risk adjustment had a strong association with hospitalization after admission to HH. Thus, it is unclear what the relationship between the average length of stay at an agency and adjusted risk of hospitalization means because length of stay was included in the calculation of expected risk of hospitalizations at an agency. Schade and Brehm found a significant positive bivariate correlation between agency average length of stay and for-profit status as well as between adjusted risk of hospitalization and for-profit status.²⁰ Given that agency average length of stay and for-profit status were correlated, does agency average length of stay—with the inclusion of individual patients' length of stay in the risk-adjustment model—represent an agency characteristic (such as quality of care differentiated by profit status) apart from patient risk factors and care needs? In other words, what is the meaning of an association between an agency's average length of stay and an agency's adjusted risk of hospitalizations when patients' length of stay has already been accounted for in calculating the expected agency prevalence of hospitalizations?

The premise that there is an average length of stay bias in the present-day CMS home health hospitalization measure rests upon the assertion that, as Schade and Brehm state it, “the probability of hospital admission increases with increasing time following hospital discharge whether or not there is intercurrent home health care” (emphasis in original).²⁰ The assumption is that agency characteristics (other than agency average length of stay) and quality of care have little if anything to do with risk of hospitalization because the risk is purportedly unrelated to whether HH is provided.

A proposed alternative to the present CMS hospitalization measure for comparison of agency performance is tracking hospital readmission rates at fixed time intervals such as at 30, 60, and 90 days from home health admission.²⁰ In simulation studies, however, it was found that simulated agency “reductions or increases in planned and observed LOS [length of stay] had no apparent impact on the fixed-interval readmission measures,” ²⁰ a finding that would seem to undermine the argument that there is a noteworthy average length of stay bias in the present CMS measure. Nonetheless, analysis of rehospitalization rates at fixed time intervals following a hospital discharge may conceivably provide information on whether or not provision of HH relates to the risk of rehospitalization; such analysis would test the proposition underlying the model outlined by Schade and Brehm²⁰ that the risk of rehospitalization relates primarily to time after a hospital discharge, not subsequent type of care provided. When HH is provided, such analysis—if it includes details on care provided—may also help reveal what attributes of HH may reduce hospitalizations.

Aside from risk adjustment for patient clinical condition, it seems that measures of the patient social environment—such as available social supports, family caregivers, and financial resources—would also need to be included in analysis of risks of hospitalizations at fixed time intervals. The patient social environment may include risk factors not within an agency's control that result in rehospitalizations within the fixed time interval but after discharge from the HHA. Without adjustment for patient environment, a fixed interval measure of hospitalization risk may be biased toward agencies that care for patients with less favorable social environments to prevent a recurrent hospitalization.

Apart from the issues regarding an agency's average length of stay, researchers have questioned the accuracy of the OASIS-based CMS hospitalization measure based on the assumption that there are reporting errors in OASIS.²⁰ In a recent study, researchers compared the measure of hospitalizations based upon OASIS with a claims-based measure of hospitalization.²² The agreement between the OASIS and claims-based measures was characterized as “quite high,” leading the researchers to state that with “the relatively good fit of the current risk-adjustment model to the claims-based measure, it is not clear that development of a new model is warranted.”²²

Information on an agency's average length of stay was not available with the Home Health Compare data in this study to further address these issues. Nonetheless, we included a number of agency characteristics as control variables including staffing levels, Medicaid utilization, and types of services provided. We also included area availability of hospital beds and state dummy variables. Schade and Brehm reported marked state-to-state variation in agency average length of stay as well as in the adjusted agency risk of hospitalization and included a dummy variable for southern states versus non-southern states.²⁰ We included dummy variables for each state minus 1 to control for state differences—differences that include, as mentioned, notable variation across states in agency-level average length of stay.²⁰ As noted in Table 2, the state dummy variables accounted for a substantial portion of the variance in risk of hospitalizations. For-profit status was associated with a greater adjusted risk of hospitalizations even after controlling for state variation (and other control variables).

In the end, the findings and discussion indicate that more research is needed to further clarify the meaning of the association of profit status, as well as other factors, with hospitalizations of patients in present-day HHAs.

Footnotes

Acknowledgments

The views presented in this article are those of the author and do not necessarily reflect the views of the government agency or its officials. The author thanks Lauren Harris–Kojetin, Ph.D., Sandra Decker, Ph.D., Jennifer Madans, Ph.D., and the reviewers of Population Health Management for their comments on earlier versions of the manuscript.

Author Disclosure Statement

The author performed this research and prepared the manuscript when employed at the National Center for Health Statistics. The author has no financial interest in any product mentioned in this article. The author's research was not supported by any commercial or corporate entity.

References

Rosenau

, Linder

. The comparative performance of for-profit and nonprofit home health care services in the US. Home Health Care Serv Q, 2001; 20:47–59.

Hillmer

, Wodchis

, Gill

, Anderson

, Rochon

. Nursing home profit status and quality of care: Is there any evidence of an association? Med Care Res Rev, 2005; 62:139–166.

Castle

, Mor

. Hospitalization of nursing home residents: A review of the literature, 1980–1995. Med Care Res Rev, 1996; 53:123–148.

Grabowski

, Stewart

, Broderick

, Coots

. Predictors of nursing home hospitalization: A review of the literature. Med Care Res Rev, 2008; 65:3–39.

Konetzka

, Spector

, Shaffer

. Effects of nursing home ownership type and resident payer source on hospitalization for suspected pneumonia. Med Care, 2004; 42:1001–1008.

Grabowski

, Huskamp

, Stevenson

, Keating

. Ownership status and home health care performance. J Aging Soc Policy, 2009; 21:130–143.

Smith

, Lapane

, Fennell

, Miller

, Mor

. Home health agency profit orientation and risk for hospitalization: A propensity score analysis of population weighted data. Home Health Care Serv Q, 2008; 27:240–257.

Choi

, Davitt

. Changes in the Medicare home health care market: The impact of reimbursement policy. Med Care, 2009; 47:302–309.

National Association for Home Care and Hospice. Basic statistics about home care. http://www.nahc.org/facts/08HC_Stats.pdf. 2010 September 20.

10.

Medicare Payment Advisory Commission. Report to the Congress: Medicare Payment Policy. Washington, DC: Medicare Payment Advisory Commission, 2009.

11.

National Center for Health Statistics. 2007. National Home and Hospice Care Survey and National Home Health Aide Survey. http://www.cdc.gov/nchs/data/nhhcsd/NHHCS_NHHAS_web_documentation.pdf. 2010 September 20.

12.

Centers for Medicare and Medicaid Services. Home Health Compare. http://www.medicare.gov/HHCompare. 2010 September 20.

13.

Area Resource File [database] Rockville, MD: Bureau of Health Professions, National Center for Health Workforce Analysis, Health Resources and Services Administration, 2008.

14.

National Quality Forum. National voluntary consensus standards for home health care: A consensus report. Washington, DC: National Qualtiy Forum, 2005. http://www.qualityforum.org/Publications/2005/10/National_Voluntary_Consensus_Standards_for_Home_Health_Care.aspx. 2010 September 20.

15.

Centers for Medicare and Medicaid Services. Home health agency risk-adjusted and descriptive outcome reports: Description and definitions. Baltimore, MD: Centers for Medicare and Medicaid Services, 2005. http://www.cms.hhs.gov/HomeHealthQualityInits/downloads/HHQIOBQIMeasureCalculation.pdf. 2010 September 20.

16.

Nuccio

, Goodrich

, Hittle

. Documentation of prediction models used for risk adjustment of home health compare OBQI quality measures. Aurora, CO: Division of Health Care Policy and Research, University of Colorado, 2008. http://www.cms.hhs.gov/HomeHealthQualityInits/downloads/HHQIHomeHealthCompareRiskModels.pdf. 2010 September 20.

17.

Harris

. An elaboration of the relationship between general hospital bed supply and general hospital utilization. J Health Soc Behav, 1975; 16:163–172.

18.

Research Triangle Institute. SUDDAN Language Manual, Release 9.0. Research Triangle Park, NC: RTI International, 2004.

19.

Institute of Medicine. Quality of Care in Nursing Homes. Washington, DC: National Academy Press, 2001.

20.

Schade

, Brehm

. Improving the home health acute-care hospitalization quality measure. Health Serv Res, 2010; 45:712–726.

21.

Shaughnessy

, Hittle

. Overview of risk adjustment and outcome measures for home health agency OBQI reports: Highlights of current approaches and outline of planned enhancements. Denver, CO: Center for Health Services Research, University of Colorado Health Sciences Center, 2002. http://www.cms.gov/HomeHealthQualityInits/downloads/HHQIRiskAdj.pdf. 2010 September 20.

22.

Williams

, White

, Bell

. Analysis of a claims-based hospitalization measure for home health. Presented at: Academy Health Annual Research Meeting, June 27–29, 2010, Boston, MA http://www.academyhealth.org/files/ARM/PodiumPresentationsCallforPapers.Pdf. 2010 September 20.