Clinical-radiological evaluation of poststernotomy wound infection

The presumption that computed tomography is the 'gold standard' imaging method for diagnosing poststernotomy sternal wound infection was never validated. This study was designed to evaluate the accuracy and role of computed tomography in diagnosing the extent of infectious complications following sternotomy. A high postoperative infection recurrence rate in our earliest cases (30 percent, 1984 to 1988) motivated us to assess whether this modality enables the surgeon to choose the optimal surgical approach, which will make it possible to reduce morbidity and mortality rates. Two-hundred three patients with poststernotomy sternal wound infections were operated upon between 1984 and 1993. All pertinent clinical and radiological data of these patients were collected retrospectively and reinterpreted by an unbiased radiologist; the radiological data were correlated both to the intraoperative clinical findings and to histological interpretation of the surgical specimens. The study group available for statistical analysis included 160 patients. Predictive statistical analysis confirmed that computed tomography is a highly reliable imaging method for identifying the different pathologies as soft tissue, sternum mediastinal infections, in sternal wound infection with overall sensitivity of 93.5 percent and specificity of 81.7 percent. New radiographic findings were identified for the distinction of costochondral infection. This complication was, and still is, a major deceptive clinical problem in these patients and the major contributor to recurrences. We propose a sternal wound infection classification system that outlines the recommended approach for each clinical-radiological condition. Since computerized tomography was found to be a highly accurate modality, we strongly believe that the surgeon should take its pathological-radiographic findings into serious consideration, even if there are no 'clear-cut' clinical signs for an existing or recurring infection.