Diagnosis-related groups using data from the National Hospital Discharge Survey: United States, 1982.

This report presents selected estimates for 1982 of diagnosis-related groups (DRG) based on data from the Na­ tional Hospital Discharge Survey (NHDS). A similar report as been published for 1981.1 The current plan is to publish orts on DRGs regularly because they determine the reim­ rsement rates of Medicare inpatients, about 30 percent of all discharges from shofi-stay non-Federal hospitals.z Developed at the Yale School of Organization and Man­ agement, DRG’s are being used by the Health Care Financing Administration, some States, and some third party payers to reimburse hospitals for inpatient care on a prospective basis.3 This approach to health care reimbursement operates on the principle that patients with similar medical conditions should receive similar care and use approximately the same amount of resources; therefore, in general a hospital should be reimbursed the same amount for each patient in a DRG. While there is variation h resource consumption among individuals within a DRG, these are expected to balance across all patients. DRG’s were developed under the guiding principle that “The primary objective in the construction of DRGs was a definition of case type, each of which could be expected to receive similar outputs or services from a hospital.”4 Their formulation was accomplished using clinical judgment and statistical procedures that classifi patients by measuring resource utilization. The first step in this process was to cluster the universe of medical diagnoses into broad, mutually exclu­ sive categories. These groups were formed to be consistent in their anatomical or physiopathological classification, or in a manner in which they are clinically managed. Once these ajor diagnostic groups were formed, an interactive statistical gram (AUTOGROUP4) was used to fimther classify each ajor group into discrete DRG’s. This process incorporated patient information regarding diagnoses (primary and secondary), procedures, sex, and age to explain maximally a patient’s length of stay. In all, there currently are 470 DRG’s. Prospective reimbursement was authorized under the Tax Equity and Fiscal Responsibility Act of 1983. Under this act hospitals participating in the Medicare program were brought into this system beginning with their fiscal year as of October 1, 1983. The Health Care Financing Administration, which operates the Medicare program, is allowing several years for hospitals to make a transition to prospective reimbursement by adjusting DRG payments based on certain hospital character­ istics and geographic location. At the end of this phase-in period, care provided to a Medicare inpatient will translate into a preestablished payment based solely on the patient’s DRG. There is an important issue related to the NHDS and the implementation of this prospective reimbursement system how this system may affect the selection of a patient’s principal di­ agnosis andlor comorbidities. Because the NHDS is designed to collect data on the morbidity of the hospital inpatient popula­ tion, any external influence on diagnostic practices may affect NHDS data. For example, two patients admitted to a hospital for treatment of chest pain-one diagnosed as having chest pain and the other diagnosed as having angina-will be placed into different DRG’s and have different reimbursement rates. There is speculation that in cases such as this prospective rehn­ bursement may influence the selection of a diagnosis,3 which in turn may affect estimates produced ffom the NHDS. After prospective payment has been in place for a few years, it may be possible to examine trends in NHDS data and determine the magnitude, if any, of this type of effect. The statistics in this report are based on data collected by means of the NHDS, a continuous survey conducted by the National Center for Health Statistics since 1965. Data for this survey are sampled from shofi-stay non-Federal general and specialty hospitals in the 50 States and the District of Columbia. The sample for 1982 contained approximately 214,000 medical