Design
Randomized, prospective, unblinded clinical comparison of 2 treatment arms with respect to intent to treat and clinical and microbiologically evaluable patients.

Setting
Six university surgical services in teaching hospitals with modern and well-staffed intensive care units.

Interventions
The consistency and objectivity of the diagnosis of pneumonia was improved by the use of a grid of diagnostic parameters. Aggressive mechanical approaches to pneumonia in intubated surgical patients were supplemented by therapeutic use of aztreonam and vancomycin hydrochloride or combined imipenem and cilastatin sodium.

Results
Patients randomized to the aztreonamvancomycin group were somewhat more ill, fared slightly better, and had fewer serious drug-related side effects than did those treated with imipenem-cilastatin (all P>.05). Immunologic parameters assessed by evaluation of bronchoalveolar lavage fluid showed differences between infected pulmonary lobes and noninfected ones; some changes were also noted in patients who recovered compared with those whose pneumonia persisted or recurred.

Conclusions
Clinical studies of pneumonia in surgical patients need to be stratified to assure comparability, to identify patients in whom treatment is likely to fail, and to display differences between more and less effective therapies. Studies of blood and bronchoalveolar lavage samples showed that certain local and systemic immunologic parameters correlate with clinical status and outcome.

Arch Surg. 1997;132:1086-1092