Data Expo 1997

The Hospital Report Cards Data

Hospitals are complex institutions, serving diverse patient populations in a variety of ways. Because of this complexity, a prospective patient is confronted with a substantial challenge in choosing the best hospital to meet his or her set of needs. A prospective patient's choice might be influenced by the advice of friends and family or by the hospital advertisements as much by the recommendations received from health care professionals, including advise from their own physician. While physicians may be better informed for making this choice, their advise, and the patient's deliberations, can be tempered by realities imposed by contractual arrangements between payer organizations, such as HMOs, and hospitals.

Growing awareness and interest in this problem has led to studies and reports by governmental agencies, research institutions, and the media. For example, with help from the National Opinion Research Center (NORC), US News and World Report has published a national ranking of hospitals in 16 speciality areas (America's Best Hospitals, John Wiley, 1996). Moreover, several governmental agencies (eg. the state health departments in Missouri and New Jersey) and private organizations (eg. the Cleveland Clinic) have published what are termed hospital report cards for the benefit of consumers faced with the hospital choice problem.

The hospital report cards data has been prepared from state data sources to serve as a part of St. Anthony's Press/Quadramed's publication entitled The Guide to Benchmarking Hospital Values. This dataset is designed to provide the raw material for constructing hospital report cards: For each of 1082 US hospitals in 10 states, data are given on 30 patient study populations (eg. obstetrics patients, pediatrics patients, cardiovascular patients, etc.) For each of these patient populations the data includes several variables (eg. number of cases, outcome rates, average length of stay, average cost, etc.)

The Problem Posed

The problem posed is to use the provided dataset, statistical methods, and statistical graphics to design a prototype hospital report card that can be used by prospective patients to select a hospital to best meet their needs.

Submissions should:

  • Address the two aspects of this problem: the practical data analysis and the presentation aspects of report card preparation.
  • Recognize the differing needs of consumers.
  • Recognize that fine distinctions (rankings) among hospitals may not be possible or reliable.

NOTE: Any report card developed for public release would include much explanatory text written by medical specialists, some references to variables not included in the hospital report card dataset, and a detailed view of the sources of the data and of the assumptions underlying the choice of data. While welcome for the Data Exposition, such material is not essential or anticipated.

How to Participate

To participate in the 1997 Data Exposition, submit a contributed paper abstract by February 1, 1997 to the ASA. Use the same form as for all other submissions to the annual meetings. The abstract form will contain a special box which should be checked to indicate that the paper is a Data Exposition submission. Please note the following:

  • A Data Exposition submission does NOT count against the limit of one contributed paper per speaker. Thus, you can contribute BOTH a contributed paper and a Data Exposition paper.
  • You must submit the registration fee along with your abstract, as for all contributed papers.
  • Your presentation will be in a Poster Session.
  • To encourage computer-based presentations, the organizers (Goodall, Cook, and Morton listed below) will welcome your correspondence on special equipment needs.

For More Information

For more information about the 1997 Data Exposition, please contact any of the following individuals: