Select for Figure 19 (11 KB), Clinical Conditions by Type of Hospital.
Select for Table 2, Ten Most Common Principal Diagnoses.
Select for Table 3, Ten Most Common Comorbidities.
Select for Table 4, Ten Most Common Principal Procedures.
Select for Table 5, Average Casemix and Length of Stay.
The results presented in this report are drawn from the Healthcare Cost and Utilization Project (HCUP), a Federal-State-Industry partnership to build a multi-State health care data system. This partnership is sponsored by the Agency for Healthcare Research and Quality (AHRQ) and is managed by staff in AHRQ's Center for Delivery, Organization, and Markets (CDOM). HCUP is based on data collected by individual State Partner organizations (including State government agencies, hospital associations, and private agencies), which then provide data to AHRQ. HCUP would not be possible without Statewide data collection projects and their partnership with AHRQ.
Statewide data organizations contribute their data to AHRQ where all data are edited and transformed into a uniform format. The uniform data in HCUP databases make possible comparative studies of health care services and the use, cost, and quality of hospital care, including:
This report is based on data from the 2003 HCUP Nationwide Inpatient Sample (NIS) which includes non-rehabilitation community hospitals (short-term, non Federal, general and specialty hospitals such as pediatric, obstetrics-gynecology, and oncology hospitals). Long-term care and psychiatric hospitals are excluded from the NIS, as are substance abuse treatment facilities. The 2003 NIS contains all discharge data from 994 hospitals located in 37 States, approximating a 20-percent stratified sample of U.S. community hospitals. The 2003 NIS includes information on nearly 8 million discharges that, when weighted, represent over 38 million inpatient hospital discharges in the United States. More information about the NIS is available on the HCUP User Support Web site at www.hcup-us.ahrq.gov/nisoverview.jsp. The 2001 and 2002 HCUP State Inpatient Databases (SID) were used to obtain prior year data on the percentage of uninsured treated by each of the hospitals in the 2003 NIS. The SID contain discharge data for all hospitals in each of the states and is the source of data for the NIS (the NIS is a sample of the hospitals in the SID).
The analyses for this report include data from two other sources of information on hospitals. The NIS was linked to data from the American Hospital Association's Annual Survey Database to obtain hospital characteristics and to the Medicare Hospital Cost Reports for 2002 and 2003 to allow for financial comparisons of hospitals.
Many definitions of safety-net hospitals exist. For example, the Institute of Medicine defines safety-net hospitals as "those providers that organize and deliver a significant level of health care and other related services to uninsured, Medicaid, and other vulnerable patients."2 Similarly, Baxter and Mechanic define safety-net hospitals as “the institutions, programs, and professionals devoting substantial resources to serving the uninsured or socially disadvantaged.”5 Other researchers have defined safety-net hospitals as those hospitals in which at least 10 percent of the costs of care provided is uncompensated. One feature that consistently defines safety-net hospitals is that they provide care to a relatively large proportion of uninsured or socially disadvantaged individuals. The definition used in this Fact Book incorporates the IOM definition, with two key modifications:
As Bazzoli and colleagues note, creating these "hard and fast boundaries" between safety-net hospitals and non-safety-net hospitals may not reflect the reality and diversity of hospitals' care for the uninsured.9 The operational definition of safety-net hospitals used in this Fact Book takes into account the actual care provision experiences of hospitals relative to one another in terms of the proportion of hospital stays for the uninsured.
Since each hospital discharge has an expected primary payer associated with it, calculating the proportion of a hospital's discharges that is uninsured is straightforward. The main HCUP categories of primary payer are Medicare; Medicaid; private insurance, including HMO; selfpay; no charge, and other. Discharges in which the primary payer was self-pay or no charge were categorized as "uninsured."
The hospitals in the 2003 NIS were divided into 10 equal-sized groups (deciles) based on the proportion of their discharges that were uninsured over a period of up to 3 years covering 2001 to 2003. Hospitals in the top decile had the highest proportion of uninsured and were classified as safety-net hospitals. Hospitals in the two deciles below the top decile were classified as secondary safety-net hospitals. To calculate the proportion of uninsured for each hospital over the 3-year period, the hospitals sampled for the 2003 NIS were matched to their records in the 2001 and 2002 State Inpatient Databases. For each hospital, the number of uninsured discharges over the 3-year period was calculated and divided by the total number of discharges over that period. For some hospitals, data were not available for all 3 years, in which case data for the available years was used. Data were available for only 2 years (2002 and 2003) for 105 hospitals (6 percent) and 1 year (2003) for 41 hospitals (4 percent).
For the analyses of clinical characteristics of patients, five broad categories that describe a patient's condition were created based on patient diagnosis-related group (DRG) in the NIS: obstetric/neonatal care, basic medical care, complex medical care, general surgery, and special surgery. This approach was an adaptation of the method developed by Stensland and colleagues10 for analyses of rural hospital care. The Stensland et al. classification was updated to incorporate recent changes to DRGs and clinically reviewed and slightly modified for applicability to general studies of community hospitals. The obstetric/neonates grouping includes DRGs related to births and newborns, including cesarean and vaginal births. The DRG system distinguishes medical from surgical DRGs. Basic medical admissions are those medical DRGs (excluding the obstetric/neonate DRGs) that would be appropriate for treatment by primary care physicians. The remaining medical DRGs (excluding obstetric/neonate DRGs) are classified as complex medical. The general surgery conditions are those surgical DRGs (excluding obstetric/neonate DRGs) that would generally be performed by a general surgeon. The remaining surgical DRGs are classified as special surgery.
Casemix index is based on the relative DRG weights provided by the Centers for Medicare and Medicaid Services (CMS). The DRG weights are a measure of the relative costliness of each DRG across all hospitals. A hospital's casemix index represents the average DRG relative weight for that hospital. It is calculated by summing the DRG weights for all discharges during the year for each hospital using DRG information in the NIS and dividing by the number of discharges for each hospital. The higher the average weight, the more resources are needed, on average, to provide care for a hospital's patients.
Several other important variables were used in these analyses. Most of these were derived from the American Hospital Association's Annual Survey Database. For example:
Selected data elements from the Medicare Cost Reports for 2002 and 2003 were added to each hospital's record and used for this report. Specifically, data on patient revenue, other revenues, operating expenses and other expenses for each hospital were added to the data file. For each year, 2002 and 2003, these values were missing for about one-third of the hospitals. Consequently, to maximize the information available for each hospital, the 2002 value for hospitals that have missing values in 2003 were used. The 2003 values for hospitals that have missing values in 2002 were used, and the 2002 and 2003 values for hospitals that have non-missing values in both years were averaged. Patient revenue margin represents the net patient revenue (i.e., patient revenue minus operating costs) divided by the operating cost of a hospital. Total income margin for a hospital is equal to the total income (i.e., net patient revenue plus contributions, government appropriations, and other income) divided by the total expenses (i.e., operating costs and other expenses).
Differences that are described in the text exhibit at least a 10-percent difference and are statistically different from zero at the 5 percent significance level (p<.05).
1. Rhoades JA. The Uninsured in America, 1996-2005: Estimates for the U.S. Civilian Noninstitutionalized Population under Age 65. Medical Expenditure Panel Survey Statistical Brief #130. Rockville, MD: Agency for Healthcare Research and Quality; 2006. http://www.meps.ahrq.gov/mepsweb/data_files/publications/st130/stat130.pdf (PDF Help).
2. Institute of Medicine. America's Health Care Safety Net: Intact but Endangered. Washington, DC: National Academy Press; 2000.
3. Institute of Medicine. Care Without Coverage: Too Little, Too Late. Washington, DC: National Academy Press; 2002.
4. Institute of Medicine. Insuring America's Health: Principles and Recommendations. Washington, DC: National Academies Press; 2004.
5. Baxter RJ, Mechanic RE. The status of local health care safety nets. Health Affairs (Millwood) 1997;16(4):7-23.
6.Holahan J, Spillman B. Health Care Access for Uninsured Adults: A Strong Safety Net is Not the Same as Insurance. Series B, No. B-42. Washington, DC: The Urban Institute; 2002.
7. Spillman BC, Zuckerman S, Garrett B. Does the Health Care Safety Net Narrow the Access Gap—Discussion Paper 03-02. Washington, DC: The Urban Institute; 2003.
8. American Hospital Association. Uncompensated hospital care cost fact sheet; 2005. http://www.aha.org/aha/content/2005/pdf/0511UncompensatedCareFactSheet.pdf
9. Bazzoli GJ, Manheim LM, Waters TM. U.S. hospital industry restructuring and the hospital safety net. Inquiry 2003 Spring;40(1):6-24.
10. Stensland J, Brasure M, Moscovice I, Radcliff T. 2002. The Financial Incentives for Rural Hospitals to Expand the Scope of Their Services. Minneapolis, MN: University of Minnesota Rural Health Research Center; 2002. Working Paper 40.
More information regarding HCUP data, software tools, and reports can be found at www.ahrq.gov/data/hcup, as well as on the HCUP User Support Web site at www.hcup-us.ahrq.gov.
Additional descriptive statistics can be viewed through HCUPnet (http://hcup.hcupnet.ahrq.gov), a free, online query system based on HCUP data.
NIS data are available for the following data years:
NIS data for years 1988 through 2004 can be purchased for research through the HCUP Central Distributor sponsored by AHRQ: Social & Scientific Systems, Inc., phone: 866-556-4287 (toll-free), fax: 301-628-3201 or E-mail: hcup@s-3.com.
Price of the NIS data is $322 for Release 1; $160 per year for 1993 to 1999; and $200 per year for 2000 to 2004. All prices may be higher for customers outside the United States, Canada, and Mexico.
AHRQ is always looking for ways in which AHRQ-funded research, products, and tools have changed people's lives, influenced clinical practice, improved policies, and affected patient outcomes. Impact case studies describe AHRQ research findings in action. These case studies have been used in testimony, budget documents, and speeches. If you are aware of any impact AHRQ-funded research or products, such as HCUP, has had on health care policy, clinical practice, or patient outcomes, please let us know using the contact information below:
Healthcare Cost and Utilization Project (HCUP)
Center for Delivery, Organization, and Markets
Agency for Healthcare Research and Quality
Phone: 866-290-HCUP (866-290-4287)
E-mail: hcup@ahrq.gov
Thanks to Dian Zheng and Jim Blakley at Medstat for developing the analytic file and to Gail Eisen, Meme Barrett, Marguerite Barrett, Craig Hunter, Nancy Jordan, David Adamson, Angela Fulmer, Katheryn Ryan, and Chaya Merrill also at Medstat, for their editorial assistance; to Margaret McNamara and Carol Stocks of AHRQ for contributions to early design decisions; to Jeffery Stensland and Julie Schoenman, currently with NORC, for sharing their system for classifying DRGs into broad service groups; to DonnaRae Castillo of AHRQ for copy editing; and to Madison Design Group for their assistance in design and layout of this Fact Book. Special thanks to Gloria Bazzoli of Virginia Commonwealth University and to Ernest Moy and Jeffrey Rhoades of AHRQ for their helpful comments on an earlier draft of this Fact Book.
Safety-Net Monitoring Initiative. Jointly led by the Agency for Healthcare Research and Quality (AHRQ) and the Health Resources and Services Administration, the key products of this initiative are a tool kit and two data books (with information at the county and metropolitan levels) designed to help policy analysts and planners at State and local levels assess the performance and needs of their local safety nets. Another publication, Developing Data-Driven Capabilities to Support Policymaking, provides guidance for using data to support the process of developing policy options for the health care safety net. (http://www.ahrq.gov/data/safetynet/netfact.htm)
Improving Efficiency through Hospital Redesign. A large integrated safety-net health care system used process flow analyses, employee focus groups, and patient and family surveys to reorganize hospital care (e.g., food service, phlebotomy, radiology, pediatrics, obstetrics) to increase efficiency. The resulting publication, A Toolkit for Redesign in Health Care, provides a roadmap and tools for redesign and improvement by other health care organizations. (http://www.ahrq.gov/qual/toolkit/)
Health Care Market and Vulnerable Populations. Researchers discovered that changes in the health care market have not eroded the safety net in the 1990s, but that an economic downturn and pressures on state budgets could mean that safety-net providers may not be able to continue to care for vulnerable populations. (http://www.ahrq.gov/research/aug04/0804RA25.htm)
Hospital Industry Restructuring: Impact On Safety Net. Investigators found that safety-net hospitals' participation in networks and systems was more common when hospitals faced less market pressure and where only a limited number of unaffiliated hospitals remained. (http://www.ahrq.gov/research/oct03/1003RA36.htm)
Child Health Insurance Research Initiative (CHIRI™). CHIRI™ consists of nine studies of public child health insurance programs and health care delivery systems that include analysis of uninsured children's access to health care for low-income children. One CHIRI™ study, Impact of Publicly Funded Programs on Child Safety Nets, found a shift in centers' clients from uninsured to Medicaid in markets with high enrollment in the State Children's Health Insurance Program, providing evidence that outreach programs for the State have a spill-over effect of enrolling previously uninsured community health center clients into Medicaid. (http://www.ahrq.gov/chiri/)
Managed Care and Community Health Centers. Researchers found that community health centers involved in managed care served a significantly smaller proportion of uninsured patients than centers not involved in managed care, and as non-managed care centers became involved in managed care, the proportion of uninsured patients they treated declined. (http://www.ahrq.gov/research/apr01/401RA17.htm)
Community Health Center Network. This practice-based research network of community health centers serves 60,000 uninsured and Medicaid managed care patients integrated computerized clinical data from different sources, created disease registries, and planned intervention research to use the diabetes registry. (http://www.gold.ahrq.gov/GrantDetails.cfm?GrantNumber=R21%20HS13543)
Evaluations of Health Disparities Collaboratives. These groups of community health centers, supported by the Health Resources and Services Administration, are engaged in rapid quality improvement of chronic care to reduce health disparities. Two evaluations are investigating the effectiveness, cost-effectiveness, and sustainability of the Health Disparities Collaboratives, as well as identifying characteristics of successful collaboratives. (http://www.gold.ahrq.gov/GrantDetails.cfm?GrantNumber=U01%20HS13635)
AHRQ Publication No. 07-0006
Current as of January 2007
Internet Citation:
Andrews RM, Stull DE, Fraser I, Friedman B, Houchens RL. Serving the Uninsured: Safety-Net Hospitals, 2003—HCUP Fact Book No. 8. AHRQ Publication No. 07-0006, January 2007. Agency for Healthcare Research and Quality, Rockville, MD. http://www.ahrq.gov/data/hcup/factbk8/
Return to HCUP
Data & Surveys
AHRQ Home Page
Department of Health and Human Services