A simple index to estimate the likelihood of bacterial infection in patients developing fever after abdominal surgery.

To identify predictors of bacterial infection in patients developing postoperative fever, we prospectively studied 434 adults who underwent abdominal surgery. Of the 434 study patients, 163 (38%) developed postoperative fever (38.1 degrees C [100.6 degrees F] or greater) and 26 (16%) of the febrile patients were found to have bacterial infection. Logistic-regression analysis showed that postoperative infection was associated with a WBC count of less than 5000 or greater than 10,000/mm3, a BUN of 15 mg/dl or greater, and fever onset after the second postoperative day. A predictive index, constructed from these three features, created a useful gradient for estimating the likelihood of postoperative infection. In patients with zero, one, two or three of the index features, the proportions having infection were 2 per cent (1/50), 14 per cent (12/88), 45 per cent (10/22), and 100 per cent (3/3), respectively (P less than 0.0001). This simple index, which uses readily available clinical data, may help reduce the cost of postoperative care by identifying patients with a low probability of infection in whom cultures, imaging studies, and empirical antibiotics do not appear necessary. Thorough diagnostic evaluation in patients with two or more index features may also reduce delay in the detection and treatment of postoperative infection. The predictive value of this index should be validated in a new patient set, however, before widespread application of the index is warranted.