This report is designed to be a tool for analysis of health care issues, and includes a wide range of data for applications by many user groups. Consumers, employers, and payers, policy-makers, and providers may begin to use this type of data to make health care decisions. Health care reform policies rely heavily on the use of objective, comparable information to drive decision-making by all parties. Utilization The tables display indicators of patient characteristics, services, resource utilization, patient case mix, source and type of admission and discharge, patient charges, payer mix, patient origins by local health district and comparative norms. It presents many of the factors within a hospital which drive the costs of patient care. The major issues addressed by these documents include: 1. volume and intensity of inpatient health care, 2. differences in inpatient services, 3. differences in patient demographics and complexity among hospitals. Consumers, employers, payers, policy-makers, and providers can utilize these documents to plan for resource allocation, identify geographic areas of public concern, weigh purchasing decisions, and make peer comparisons. Purchasers may use the information to select providers and payers, or to tailor benefit offerings. Users of this report must remember that several factors such as volume of patients discharged, coding inconsistencies, and severity of patient illness can influence inter-hospital comparisons. In interpreting the information shown in this report, the reader is advised to keep in mind the following: Volume If a hospital discharged only a few of a certain type of cases, comparing data with other hospitals would not be especially meaningful because a small number of cases is not sufficient to establish a pattern of treatment. The reader must exercise caution when interpreting measures shown in this report that were based on less than five discharges. Coding From its beginning, the committee worked to assure the best data quality possible. To do so, they implemented the following: 1. The Health Data Plan provides data element definitions and standards to ensure all hospitals will report similar data. 2. Systematic edits were put in place to identify missing or invalid data fields and hospitals are required to correct these. 3. All discharge records are subjected to a second round of editing which checks for potential problems in a record related to highly improbable clinical situations. (A detailed description of systematic edits and clinical coding edits are included in the DATA PROCESSING AND QUALITY section below.) 4. Each hospital is provided with two 35 day review periods to validate the committee's data against their hospital records. Despite the detailed edit and validation process, data quality is still an issue but is expected to improve over time as hospitals become accustomed to reporting data for public dissemination. The committee is working with individual hospitals to improve data quality by comparing their coding error rates with state norms. Any comparative analysis, or decision-making, based on these data, should take into account issues of data quality. Severity of Illness Patients entering hospitals for the same treatment and conditions often vary in the severity of their illnesses. Factors such as age, gender, and secondary illnesses account for differences in severity. Treating severely-ill patients is the most resource intensive and expensive. For instance, patients who are the sickest may need to be admitted to intensive care units, may need high-technology equipment, or may need to stay longer in hospitals than those less ill patients. Some hospitals, especially regional referral centers such as Primary Children's Hospital and LDS Hospital, treat more acutely ill patients because of the specialized care available at their facility. The University of Utah hospital, which serves as a regional referral center as well as a major teaching hospital, treats more patients with complex medical conditions than other hospitals. Charges for patients cared for at these hospitals may be higher than at other hospitals due to the type of services offered and the type of patients served. Rural hospitals often admit a mix of patients that may be chronically ill, uninsured, or elderly. The elderly are often more severely ill because of chronic and multiple health problems.
DIAGNOSIS RELATED GROUP (DRG)The DRGs were developed for the Health Care Financing Administration as a patient classification scheme which provides a means of relating the type of patients a hospital treats (i.e., its case mix) to the costs incurred by the hospital. While all patients are unique, groups of patients have common demographic, diagnostic and therapeutic attributes that determine their resource needs. All patient classification schemes capitalize on these commonalities and utilize the same principle of grouping patients by common characteristics. The use of DRGs as the basic unit of payment for Medicare patients represents a recognition of the fundamental role a hospital's "sicker" patients play in determining resource usage and costs, at least on average. "The DRGs, as they are now defined, form a manageable, clinically coherent set of patient classes that relate a hospital's case mix to the resource demands and associated costs experienced by the hospital." (Diagnosis Related Groups, Seventh Rev., Definitions Manual, page 15.) Each discharge in the UHDDB was assigned into a DRG based on the principal diagnosis, secondary diagnoses, surgical procedures, age, sex, and discharge status of the patient. In the 1992 UHDDB, each patient is assigned into one of 492 DRGs.
ALL-PATIENT REFINED (APR) DRGThe APR-DRGs are a patient classification scheme developed by 3M Health Information Systems (HIS) that follows the basic DRG methodology of classifying patients into disease categories, but further subdivides each disease category into severity of illness classifications. With a few exceptions, a patient in each disease category (called consolidated DRG) is assigned into one of four levels of severity: no/minor complication or comorbidity (CC), moderate CC, major CC and extreme CC. Some of the exceptions to the four-level classification are newborns and neonates which are assigned to APRDRGs formed with the severity of condition already built-in (e.g., APRDRG 606: Neonate, birthweight 1000-1499g with significant O.R. procedure, discharged alive). APRDRG categories do not appear in this report, but were used to define charge and length of stay outliers and calculate the Case Mix Index (See CASE MIX INDEX). The Health Data Committee has published a separate report at the APR-DRG level, showing average charges and length of stay adjusted for severity level. The report, entitled "SP-1 Patient Severity, Total Charges and Length of Stay" was done for 1992 discharges.
DATA PROCESSING AND QUALITYData Submission The UHDC receives discharge data quarterly from hospitals in various formats and media. Most of the unaffiliated small rural hospitals submit hard copies of UB-82 forms. Discharges from affiliated hospitals are submitted in electronic format by the corporate office (IHC; Health Trust, Inc.; and Holy Cross). Discharge data are converted into a standardized format by the Office of Health Data Analysis as specified in the Health Data Plan. System Edits Data are validated through a process of automated editing and report verification. Each record is subjected to a series of edits that check for accuracy, consistency, completeness, and conformity with the definitions specified in the Technical Manual. Records failing the edit check are returned to the data supplier for correction or comment. Clinical Claims Edit All discharge records are subjected to a second round of editing using 3M HIS Clinical Claims Edit (CCE) software. The CCE flags records when any of 25 edit conditions are detected. Table 1 summarizes the conditions which may result in the record failing the edit process and requiring a correction or explanation from the hospital. Table 1 Conditions Flagged as Possible Errors by CCE ____________________________________ 1. Procedure unlikely with diagnosis 2. O.R. procedures coded are not usually performed for principal diagnosis 3. Principal diagnosis suggests surgery but no O.R. surgery performed 4. Symptom code as principal diagnosis 5. Clinically unreasonable length-of-stay (high or low) 6. Questionable admission 7. Age conflict 8. Sex conflict 9. E-Code as principal diagnosis 10. Manifestation code as principal diagnosis 11. Non-specific principal diagnosis 12. Open biopsy check 13. Unacceptable principal diagnosis 14. Non-specific O.R. procedure 15. Duplicate of principal diagnosis 16. Bilateral procedure 17. Invalid diagnosis or procedure code 18. Invalid 4th or 5th digit 19. Duplicate code 20. Evaluate as principal diagnosis 21. Requires secondary diagnosis 22. Diagnosis conflict 23. Procedure conflict 24. Maternal/Newborn code conflict 25. Invalid or unknown age _____________________________________ Appendix A contains a description of each of the error flags listed below.
OUTLIER CASESSome patients have exceptionally low or high lengths of stay (LOS) or total charges. A hospital's charges can be affected by just a few unusually long (or short) or expensive (or inexpensive) cases. These high or low values could be a result of coding or data submittal errors, particularly in length of stay, total charges, or data elements that affect DRG assignments. Other reasons for an exceptionally low LOS or charges could be due to death or transfer to another facility. Exceptionally high LOS or charges could be due to a catastrophic condition. Whatever the reason, these values, referred to as "outliers", distort the averages and were excluded from calculations. LOS or charge high outliers are defined in this and succeeding reports as values above 2.5 standard deviations from the mean. Means and standard deviations are APR-DRG specific and calculated on a statewide basis. The low outliers were defined as a non-newborn or non normal delivery discharge with less than $300 charge. However, the calculations in this report do not exclude low outliers. A preliminary analysis showed that of the 449 discharges that met this definition, a high proportion are in the DRG "Other factors influencing health status", for which it was difficult to determine whether they were true outliers.