The case against using the APACHE system to predict intensive care unit outcome in trauma patients

The use of outcome indices as a means of evaluating institutional performance for delivery of medical care is at the forefront of federal health policy reforms. Because an enormous number of clinical and financial data are generated by ICU patients, it is inevitable that integrated bedside computers will be necessary to supply the type of information that is being sought by governmental and private insurance agencies involved in assessment of hospital performance. The Health Care Financing Administration already has adopted the APACHE data collection protocols and predictive models for the severity of illness adjustments that were used in assessing the 1986 hospital-specific death rate for acute myocardial infarction, congestive heart failure, stroke, and pneumonia. In our opinion, however, it is unlikely that any single system will be developed that can accurately estimate more than 50% of ICU deaths. The intention of the APACHE III system to include 78 diagnostic categories seems unrealistic. Furthermore, the number of data needed to document outcomes for both low- and high-risk admissions is impractical. We are evaluating APACHE III to determine whether the revisions to the definition for head trauma will represent a significant improvement in predicting outcomes for trauma patients. In the interim, the financial investment in the APACHE III automated bedside data collection system cannot be justified for trauma patients. Neither should it be used in ICUs that admit a large number of trauma patients as a tool for monitoring unit efficiency, guiding triage decisions, allocating staff and ICU beds, identifying risks of iatrogenic or other potential complications, or assessing quality of life, in spite of marketing efforts by the APACHE Corporation. We believe that using any of the APACHE systems for these purposes, at best, is premature, and potentially misrepresents the trauma patient population. Standards for patient classification already are in place for use in making determinations for institutional reimbursement from governmental and insurance agencies. The inequities for certain subgroups of patients, including trauma patients, could create situations in which care is rationed rather than allocated according to a plan that distributes resources efficiently. The APACHE system has several shortcomings and adds little, if anything, to the potential solutions for trauma quality assurance and resource allocation. Nor has the APACHE system established procedures for documenting institutional review of unexpected trauma deaths that would be equivalent, for example, to the type of audit filters applied by the American College of Surgeons in conjunction with the TRISS methodology.(ABSTRACT TRUNCATED AT 400 WORDS)