A strategy for statistical Master Person Index linking.

A linking program used by Connecticut Healthcare Information Management and Exchange to maintain the Master Person Index for its large, state-wide patient data repository is being stretched beyond its limits by the growing size and complexity of the database. This paper presents the early work into developing a second-generation linking program. Like the original program, the new linker will use a unique multi-step process to allow effective linking of data from a large number of dissimilar data sources. The new linker will use parallel multi-processing to allow improved performance and scalability. These changes will also make possible more sophisticated statistical methods of defining link confidence. The system is implemented using a scalable collection of inexpensive, PC based systems running the Linux operating system, a freely available database engine, and the Java programming language.

Hospital charges for terminal care of cancer patients dying before age 65.

Cancers are common causes of death before age 65 years, but health care costs have rarely been examined. Data from the statewide population-based Connecticut Tumor Registry were linked with the statewide hospital-discharge database, for cancer patients who died in 1992 before age 65 years. Average cancer-related charges for stays at Connecticut hospitals during the last year of life for lung and breast cancer were $31,000-$37,000 per patient, and total charges were $9 million-$12 million. Trends in the health care system and in patient preferences may affect future hospital charges for terminal care of cancer patients.

Estimating breast cancer treatment charges in Connecticut.

Previous studies of estimated costs for cancer treatment have been limited to elderly patients or to specific health maintenance organizations. Data from the statewide population-based Connecticut Tumor Registry on a random sample of 407 breast cancer patients diagnosed in 1991 were linked with a statewide hospital-discharge database, to estimate charges (through September 1993) for inpatient and ambulatory surgery care. For the 377 cases (92.6% of 407) successfully linked, average charges attributed to breast cancer care declined with age, increased with extent of disease (stage at diagnosis), and increased with extent of surgery; these associations persisted in multivariate analyses. Total hospital-related charges for comorbid conditions (during 1991-93) were considerable by age 45 to 64 years. The merged database should be most useful in estimating charges for: cancers treated mainly by surgery (including ambulatory surgery at hospitals); comorbid conditions; and terminal care.

Verification of information in a large medical database using linkages with external databases.

Efforts to utilize Uniform Hospital Discharge Data Sets (UHDDS) for epidemiological studies have been hampered by the limitations of those databases. The purpose of this paper is to illustrate that linking to external databases can provide the verification necessary to overcome many of those limitations. This method has dramatically altered study design at the Connecticut Hospital Research and Education Foundation and has provided an efficient method for specifying data collection weaknesses within the resident databases.

The "Toward Excellence in Care" program: a statewide indicator project.

BACKGROUND: The "Toward Excellence in Care" program was launched by Connecticut hospitals and physicians in 1988 to develop and use quality-of-care indicators for use in quality improvement. METHODOLOGY: Data came primarily from the Connecticut Health Information Management and Exchange (CHIME) database, which contains discharge abstract information, UB (uniform billing)-82 information, and additional data elements, for all of Connecticut's 34 acute care hospitals. Linkages also occur with the state mortality database, the trauma registry, and with admission/discharge data within and across Connecticut hospitals. The "Toward Excellence in Care" program staff help the hospital use the data on indicator reports for quality improvement. EXAMPLE: On receiving a report on care for patients with acute myocardial infarction (AMI) a program representative summarized opportunities for improvement. The data were then disseminated to both the cardiology and the hospitalwide quality improvement staffs. Cardiologists conducted chart review on 100% of patients included in the last time-frame on the report (for example, fiscal year 1991). The quality improvement professional documented the system of care for an AMI patient. Recommended actions included adoption of a policy for emergency department administration of thrombolytic therapy before a cardiology consultation, and modification of the postcoronary care program. CONCLUSIONS: Progress in addressing four challenges-easing the burden of data collection on the hospitals, maximizing acceptance of information by hospitals and physicians, risk adjusting data to permit comparison of outcomes, and facilitating understanding of reports--is reflected in expanding use of the "Toward Excellence in Care" program.