Robert B. Fetter

AUTOGRP: an interactive computer system for the analysis of health care data.

AUTOGRP is an interactive computer system designed to facilitate rapid analysis of complex medical information. AUTOGRP allows the clinical or administrative expertise of the user to be combined with sophisticated computer techniques to permit rapid information retrieval, hypothesis testing, development of norms, and identification of deviant cases. This interaction yields results of a uniquely high statistical and medical quality. AUTOGRP has been used to aid in understanding the process of patient care management in a variety of settings in order to enhance the effectiveness of decision-making from both a medical and management point of view.

Diagnosis-related group refinement with diagnosis- and procedure-specific comorbidities and complications.

Diagnosis-related groups have been revised through more refined uses of secondary diagnoses. Under the refined diagnosis-related groups, patients are distinguished with respect to classes of secondary diagnoses that are disease- and procedure-specific. Each class represents a different level of utilization for a given principal diagnosis or surgical procedure. The refined system was evaluated with national data from the Medicare program. Estimates of hospital costs and utilization based on refined diagnosis-related groups were more precise than those based on unrefined diagnosis-related groups. This approach to diagnosis-related group refinement does not represent a radical departure from the current diagnosis-related group framework and does not require new data collection efforts. Moreover, a payment system based on the refined model is less affected by the ordering of the diagnoses than under the existing diagnosis-related group system. How the refined diagnosis-related group framework can accommodate future refinements at all levels of the classification scheme is also discussed.

The application of diagnostic specific cost profiles to cost and reimbursement control in hospitals

A system has been developed to generate hospital budgets based on the types of patients served. Several hundred classes of patients are defined according to clinical attributes such as diagnoses and surgical procedures, and for each class a profile of resources consumed is determined. The class definitions are based both on homogeneity of patient care processes as well as resource consumption. These profiles are expressed as revenues generated by charging departments and as costs both direct and indirect for all services. A methodology has been developed to associate all indirect costs with their source for each service included in the profile. From a forecast of patient load by class, budgets can be computed from the cost profiles and revenues determined from the charging profiles. Further analysisthus can included the effect of changes in case mix as well as changes in patient care processes. The effect on revenues of different reimbursement mechanisms can also be projected as a function of the case mix. The system is currently being implemented for demonstration and evaluation at the Yale-New Haven Hospital.

A Decision Model for the Design and Operation of a Progressive Patient Care Hospital

The problem of determining the number of beds required in each type of service in a progressive hospital, subject to constraints of operating and investment costs and minimum service levels, is approached through computer simulation and mathematical programming. The simulation of the progressive patient care system is based on patient flow through the service zones in an operating hospital, but the program is capable of accommodating any other hospitals structure of facilities or set of patient paths. The mathematical programming approach to the solution of optimal bed allocation raises basic questions on the formulation of objective function and budgeting constraints which are open to further investigation.

Case mix of public patients in skilled nursing facilities in Connecticut.

The case mix of publicly funded residents in 73 skilled nursing facilities (SNFs) in two Connecticut counties was examined. Data collected in 1980-1981 for utilization review by a professional standards review organization were used for the analysis. The findings indicate that considerable variation exists in case mix across the SNFs. Medicaid per diem rates, which are based on historic costs, have a low and negative correlation with case mix. Case mix indices are significantly higher for rural facilities, with a for-profit type of ownership, and with less than 75% of the bed days devoted to Medicaid residents.