Salmonella lung abscess in a patient with acquired immunodeficiency syndrome.

considered are extremely large) and would be of no use in medical audit. Third, we prefer to use a system for evaluation of severity of disease that analyzes the modifications of basic pathophysiologic patterns of response to disease, as the Apache II system does, because it represents in our opinion the best tool for risk evaluation in all kinds of patients. The Apache II system was applied in our work to high-risk patients only. We agree that APACHE II is not the best system to categorize the single patient admitted for scheduled surgery, but it may be a good research tool for high-risk surgical lung carcinoma patients, who represent a heterogeneous group. Finally, we think that the best results in terms of accuracy can be achieved in large series of patients.