[ Main page of Report | Contents of Report ]
In this chapter, we provide a general overview of data sources and variables used in our analyses. We used the data set to support the following analytic tasks:
Ideally, all required data would be available for all hospitals across the country for the same time period. Substantial information on individual hospital characteristics is available from the national sources discussed below. However, some utilization and financial data that are needed to measure hospital services to low-income populations (or at least the resources devoted to Medicaid patients and self-pay patients) are not directly available. Detailed inpatient utilization data on self-pay patients are available only for the 20 percent sample of hospitals from the 24 states included in the Hospital Cost and Utilization Project (HCUP) database. Much of the needed information on revenues by payer and on uncompensated care is collected in the American Hospital Association (AHA)'s Annual Survey, but confidentiality considerations preclude using that information for detailed exploratory analyses requiring hospital-level information. Thus, we supplemented the available national data with the HCUP national sample and detailed claims and financial data from three states: California, New York, and Wisconsin.
[Go To Contents]
A Medicare cost report is submitted by each Medicare participating hospital based on the hospital's own fiscal year.(1) The Centers for Medicare and Medicaid Services (CMS) makes selected data from the Hospital Cost Report Information System (HCRIS) available as a public use file. This file contains data on utilization (Medicare, Medicaid, and total), Medicare costs and payments, total costs and charges for routine, ancillary and outpatient services, and data from the hospital's financial statement. These data have been used to describe the relationship of Medicare payments and costs for categories of hospital services covered by different payment systems, hospital cost patterns and trends, and overall hospital financial status.
The data from the financial statements are generally less reliable than data elements used for payment purposes. The instructions indicate that the worksheets should be prepared from the hospital's accounting books and records and do not provide specific line item definitions. The fields are normally not audited; however, adoption of a later filing date for certified cost reports may have improved the quality of the financial data in recent years. The cost report data are updated quarterly by CMS.
The cost report has inpatient utilization statistics for Medicare and Medicaid patients (discharges, days) but not for self-pay or charity care patients. Utilization statistics are not collected for outpatient visits. There is no information on revenues by payer class (other than Medicare charges). The only cost information is for Medicare patients and all patients. No information is collected on charity care.(2)
The financial data cannot be used in its current form to determine uncompensated care costs but can be used to establish overall measures of financial viability.(3)
In addition, the cost report data can be used to develop cost-to-charge ratios for inpatient services that can be applied to hospital charges to estimate the cost of providing services.
The most recent AHA survey available for our report was for 1999. In completing the survey, hospitals are requested to report data for a full year based on their fiscal year (so that the reporting period should be consistent with Medicare cost report data). Overall, the average response rate is about 82 percent. However, the response rate on many of the financial items relevant for our purposes is about 65 percent and is not necessarily representative of the universe of hospitals (e.g. proprietary hospitals have a lower response rate for financial data). This makes it difficult to analyze data within states or market areas. Estimates are made of data for non-reporting hospitals and for reporting hospitals that submit incomplete AHA Annual Survey questionnaires. Payer mix and revenue information is confidential and is not available on the public use file. On the non-confidential file, inpatient utilization information by payer is available for Medicare and Medicaid patients but not for other payers.
CMS maintains a public use file with information from the certification system on Medicare/Medicaid participating hospitals. The file contains information on a number of hospital characteristics, including location, type, ownership, beds, and staffing. The information is updated only when actions affecting survey, compliance and certification are taken with the hospital so that variables such as staffing may be quite dated. Licensed beds (as opposed to beds that are staffed and maintained) are reported.
CMS produces the PPS impact files each year as part of the annual update in the hospital prospective payment system. The files contain information that can be used to estimate each hospital's payments for the upcoming federal fiscal year. The variables include current payment parameters (e.g. the applicable wage index for the upcoming fiscal year), hospital characteristics from the most recently filed cost report, and patient characteristics from Medicare claims data from two years prior (e.g. the FY2001 impact file is based on MEDPAR data for FY 1999). Relative to the cost report data, the information is timelier and allows a direct estimation of prospective payments (including DSH) by federal fiscal year.
The CMS intermediaries that process Medicare hospital claims maintain the provider-specific file to price claims for inpatient hospital services. A public use file is made available annually each year as part of the PPS update that contains all the payment parameters used to pay hospitals for inpatient services in the current year. The file includes certain information (e.g., how a hospital is paid under the capital prospective payment system) that is not available on the PPS impact files.
CMS maintains two public use files related to expenditures under Medicaid DSH programs.
The Area Resources File (ARF) maintained by the Health Resources and Services Administration is a prominent source of information on the area in which each hospital is located. The ARF is a computerized health resources information system that contains more than 7,000 variables at the county level that could be aggregated to other geographic groupings such as MSAs/PMSAs. The ARF integrates data from a variety of sources, including the AHA, Bureau of Census, Interstudy, CMS and the VA. The variables include information on health care facilities and professions, population characteristics and economic data, and utilization. The 2000 release was available for our analyses, which draws most data elements from 1998.
The Kaiser Family Foundation's State Health Facts Online (www.kff.org) contains state-level data on demographics, health, and health policy, including health coverage, access, financing, compiled from a variety of sources, including CMS and Current Population Surveys. It is a readily available source for information on Medicaid enrollment and expenditures and estimates of the uninsured in each state.
[Go To Contents]
The Agency for Healthcare Research and Quality's Healthcare Cost and Utilization Project (HCUP) consists of a national inpatient database containing 100 percent of the inpatient claims from a sample of hospitals in 24 states. Although the sampling frame is limited to participating states, it is selected to reflect characteristics of community hospitals nationwide. Weights are provided that can be used to produce national and regional estimates or state estimates for the participating states. We used the 1998 HCUP data in our analyses.
The HCUP files contain more than 100 clinical and non-clinical variables included in a hospital inpatient discharge abstract, including diagnostic information, patient demographics, length of stay, and--most important for this project--total charges and expected primary and secondary payment source (such as Medicare/Medicaid/private insurance/self-pay). The HCUP data can be used to develop measures of inpatient hospital utilization and case mix by payer class. The charge data can be used with cost-to-charge ratios from the Medicare cost reports to compute costs by payer category for each hospital. Payment amounts are not provided. The sampling means that the data cannot be used to develop measures of market share for specific hospitals
In addition to the national inpatient database, each participating state maintains a State Inpatient Database (SID). The SID contains data on all discharges from all community hospitals in the participating states. We used the 1997 SID for three states in our exploratory analyses: California, New York, and Wisconsin.
There is no comparable database for outpatient services that provides utilization data by payer.(4)
A number of state databases have detailed hospital-level financial data that could be used as an alternative to the AHA survey data. Some are a by-product of a state uniform reporting system with detailed definitions for each data element. The States that we are using in our analyses- California, New York, and Wisconsin- all have detailed instructions concerning how data should be reported.
[Go To Contents]
In this section, we provide an overview of the hospital variables that we constructed to support our analyses. The focus of our attention was on acute care community safety net hospitals that are financially vulnerable. Therefore, we built a detailed data set for all community hospitals (other than Maryland where an all-payer system applies), including children's and cancer hospitals. While we included other specialty hospitals and institutions for mental disease in our database to the extent they receive Medicaid DSH funds, the data for these hospitals was more limited.
We used HCFA's Provider of Service file created January 1, 1998 to establish our universe of hospitals. We identified Medicare-participating hospitals by their Medicare provider number. Hospitals that are not Medicare participating facilities (e.g., institutions for mental disease) were assigned a dummy provider number. Below, we provide a general description of the data elements we included and the sources for the data. Where the variables were created for specific analyses, more detailed information on the data and methods are provided in the chapter reporting on those analyses.
Hospital capacity and services: We used the Medicare Provider of Service file and the AHA survey public data to determine basic hospital characteristics such as type (e.g., general acute care, children's, long-term), ownership, and bed size. We used the Medicare cost report to determine the number of FTE residents and teaching intensity.
Hospital utilization: We used the Medicare cost report and/or AHA survey public data to determine inpatient days and discharges for Medicare and Medicaid patients and all patients. (Other utilization data by payer, e.g. self- pay inpatient days and outpatient visits, is not available from the AHA survey data.) We used the AHA's adjusted days as an overall measure of hospital capacity.(5)
Low-income patient measures: Development of low-income measures is hampered by lack of available data. We developed different measures based on available data that enabled us to assess the distributional impact of using different definitions of financially vulnerable safety net hospitals.
All acute care hospitals. We used the Medicare cost report data to generate Medicare and Medicaid utilization statistics for all acute care hospitals. We applied the SSI percentage to the Medicare days to estimate low-income Medicare days. We used these data in evaluating the current distribution of DSH payments (Chapter 4).
1998 HCUP national database. We constructed inpatient measures determining:
We used these data in analyses exploring the implications of using different utilization measures to allocate DSH funds (Chapter 7).
Three-state analysis. We used the databases for California, New York and Wisconsin hospitals for an in-depth analysis of the sensitivity of the allocations to using different measures (Chapter 8). The state financial data have detailed information by payer source on inpatient and outpatient gross revenues and on uncompensated care. In addition, we had access to 100% of their 1997 inpatient claims through HCUP. Our measures from the financial statements included:
We used Medicare cost reports (HCRIS 13, 14 and 15) to generate a standard set of financial measures for all hospitals in the database covering the four financial domains: profitability, liquidity, capital structure, and asset efficiency (Chapter 5).
[Go To Contents]
1. Medicare allows low-Medicare utilization hospitals such as children's hospitals to file low-volume reports that do not contain all the cost information required for the full cost report.
2. Under Medicare rules, bad debts, charity, and courtesy allowances are deductions from revenue and are not an allowable costs; however, bad debts attributable to the deductibles and coinsurance amounts for Medicare beneficiaries are reimbursable at 70% (effective in 2001 as a result of the BIPA restoration of BBA reductions).
3. In the future, data on uncompensated care costs should be available through the Medicare cost report. The BBRA requires the Secretary to collect through the cost report data on costs incurred by acute care hospitals in providing inpatient and outpatient hospital services for which the hospital is not compensated, including non-Medicare bad debt, charity care, and charges for Medicaid and indigent care. The provision is effective for cost reporting periods beginning on or after October 1, 2001. HCFA has not issued final instructions implementing this provision.
4. Nine states provide information on ambulatory surgery that includes payer information.
5. The adjusted days reflect the number of days of inpatient care, plus an estimate of the volume of outpatient services, expressed in units equivalent to an inpatient day in terms of level of cost. Outpatient services are converted to the equivalent number of inpatient days by multiplying the number of outpatient visits by the ratio of outpatient revenue per outpatient visit to inpatient revenue per inpatient day.
Main Page of Report | Contents of Report
Home Pages:
Office of Health Policy (HP)
Assistant Secretary for Planning and Evaluation
(ASPE)
U.S. Department of Health and Human Services
(HHS)