Further appraisal of APACHE II limitations and potential

Hemodynamically unstable patients selected for invasive cardiovascular monitoring were divided into APACHE II subgroups for risk stratification to study interrelationships among monitoring, therapy, resulting cardiovascular function and outcome. When compared by regression analysis, there were no clinically relevant relationships between APACHE II scores and total intervention points (r2 = 0.02), days of invasive monitoring (r2 = 0.000001), initial cardiovascular function (r2 = 0.069) and final cardiovascular function (r2 = 0.05). Analysis of variance (ANOVA) was done between APACHE subgroups and total points (zero of 20 intragroup comparisons were different by the Scheffé test; p = 0.33), days of monitoring (zero of 20 were different; p = 0.61), initial cardiovascular function (three of 20 comparisons were different; p = 0.003) and final cardiovascular function (zero of 20 were different; p = 0.24). Opposite relationships in patients who lived and died were noted between total intervention points and APACHE II subgroups (p = 0.028, two-way ANOVA). There was an increasing number of total intervention points in patients who ultimately lived in ascending initial APACHE II subgroups. In contrast, there was a decreasing number of total intervention points in patients who ultimately died in the same APACHE II subgroups. APACHE II stratification failed to help understand the relationships among clinically important parameters. At the same time, while APACHE scores are claimed to be independent of therapy, the score seemed to be extremely sensitive to interventions, especially important in surgical populations. Should the APACHE II scores remain high in the face of continued maximal intervention, fatal outcome can be predicted. This pattern is remarkably similar across the entire initial APACHE spectrum. The predicated attributes of APACHE II scores, that is, risk stratification and independence from therapy, are neither necessary or desirable. Understanding patterns that are associated with survival or death may require alternative mathematic approaches, such as group and set theory manipulated by principles of Boolean algebra. New approaches may be more fruitful than further attempts to refine existing systems.