Health care transformation and the case for a community-wide health information management environment.

The transformation of the health care delivery system in local, metropolitan, and regional markets is progressing rapidly. This transformation is fueled by competition, the shift of financial risk to the provider continuum, employer demands for cost containment, and the breadth and depth of state and federal government reform initiatives. However, information systems do not yet exist to support these transformations. We propose establishing a new community-level information management environment and new measures of health care system performance.

Severity of Illness Index and the Adverse Patient Occurrence Index. A reliability study and policy implications.

Seven hundred fifty-two randomly selected charts from seven teaching hospitals were rated by pairs of medical record analysts. The Severity of Illness Index was unreliable with an interrater-agreement rate of 73% (kappa statistic = 0.41), and demonstrated a significant (P less than 0.0001) association with the Adverse Patient Occurrence (APO) Index. This suggests that the Severity of Illness Index is not differentiating severity of illness from quality of care. The fair to poor field reliability stems from underlying instrument subjectivity, lack of clear referent groups, and time pressure. The APO Index was also found to be unreliable (r = 0.33 and range = -0.05-0.58). Greater attention should be directed to improving objective discharge abstract, billing, and laboratory data for measuring patient severity and adverse patient occurrences.

Prospective payment for psychiatry--feasibility and impact.

Although 15 diagnosis-related groups (DRGs) have been proposed for psychiatric hospital patients, psychiatric hospitals are currently exempt from the DRG prospective payment system. We investigated the ability of the psychiatric DRGs to predict the hospital length of stay and costs by retrospectively analyzing the charts of 8816 randomly selected patients from 32 psychiatric hospitals throughout the United States. In addition, we developed other grouping systems to see whether they would have been better predictors of length of stay. We found that grouping the patients in the 15 psychiatric DRGs reduced the total variance in length of stay by only 3.9 percent. Furthermore, our best alternative grouping--based on major diagnostic categories, whether the patient was transferred from another facility, age, and psychiatric complications and comorbidities--reduced the variance by only 7.8 percent. We conclude that DRGs do not adequately predict length of stay or costs in psychiatric hospitals. We identified factors other than diagnosis that predicted the length of stay better, but all the models we tested would create large financial "winners" and "losers" and thus introduce inappropriate incentives into the care of patients in psychiatric hospitals.

Comparing classification methods: measurement of variations in charges, length of stay, and mortality.

We define and examine three alternative systems for categorizing hospital patients. The first system is based on discharge abstract data alone; the second system is based on discharge abstract data that has been reabstracted from charts for completeness and accuracy; the third system is based on a severity-of-illness index within a diagnostic group. Using acute myocardial infarction patients as an illustrative example, we examine the homogeneity of these categorization systems with respect to charges, length of stay and mortality rates. Our results indicate that a measure of patient severity of illness is essential to produce homogeneous categories to study hospital productivity for utilization review assessment by PSROs, for planning purposes by Health Systems Agencies, for hospital rate setting by cost review commissions and for internal hospital financial management and utilization review studies.

Medical staff and hospital administration uses of discharge abstract data.

In summary, this article suggests several useful approaches whereby physicians and hospital administrators can efficiently sort through large amounts of discharge abstract data to identify problem areas, particularly those concerning quality of care and utilization of services, and to formulate strategies for improvement. Approaches and steps suggested include: . Review objectives-determining variables of greatest concern; . Standards-determining acceptable values or levels of performance; . Focus of review-determining the unit of review, such as specific departments, diseases, or physicians; and . Quality and reliability of discharge abstract data-determining the data to use by undertaking four levels of edits: technical edits, logical edits, ratio edits, and categorical edits. A model flowchart provided in this article serves as a conceptual framework by which potential problem areas can be studied during a patient's hospitalization to precisely identify the interaction, the time, and the cause/association of the problem.

Hospital cost per case: analyses using a statewide data system.

In this paper, we establish relationships between hospital cost per case and the independent variables; case mix complexity, case mix severity, factor input prices, and hospital characteristics. Two hundred and sixteen thousand discharges from Maryland's acute general hospitals are grouped into 383 Diagnostic Related Groups which are used to compute an information theoretic measure of case mix complexity. Multiple linear regression equations are developed which predict up to 88% of the variance of between-hospital cost per case. The most highly significant predictors of cost per case are complexity, patient age, proportion of high risk patients, average length of stay, and nonphysician salary levels. Two distinct groups of hospitals, metropolitan and rural, are defined and models are developed for each. We discuss the implications of these findings for the identification and regulation of unexpectedly high cost hospitals and for prospective cost per case reimbursement.

An analysis of case mix complexity using information theory and diagnostic related grouping.

Case mix complexity measurements are essential to determine health care efficiency and effectiveness. Measures of patient care processes and outcomes must be adjusted for case mix before valid comparisons can be made. Hospital reimbursement, particularly prospective reimbursement, must take into account differences in case mix. In addition, a key variable for hospital classification is case mix. There are, however, no widely accepted easily computed case mix measures. Information theory measures of case mix have been developed but their acceptance has been limited by a lack of verification of their basic assumption that concentration of disease is related to clinical complexity. We discuss the rationale underlying the mathematical computaton of information theory measures and demonstrate a statistically significant relationship between clinical measures of case mix complexity and information theory measures of case mix complexity.