Is Sternal Revision after Open Heart Surgery Predictable? Development of a Risk Assessment Score for Reoperation Caused by Sternal Wound Infection

Objectives: Despite the progress of surgical technique and postoperative wound care management, deep sternal wound infection (DSWI) is still associated with high mortality and morbidity. Diagnosis is often delayed and therapy is complicated by late and gradual onset of wound healing disturbances, even after discharge from hospital. The aim of this study was to develop a postoperative clinical tool to identify patients with high risk for sternal revision caused by sternal wound infection.

Methods: In 260 patients standardized data of clinical and wound healing parameters were prospectively collected over a 30 day period following open-heart surgery. To predict the necessity for reoperation caused by wound healing disorders these data were retrospectively analyzed. Using a receiver operating characteristic (ROC) for modeling, the risk of reoperation caused by wound healing disturbances was estimated prospectively.

Results: Overall 13 patients (5 %) suffered a reoperation caused by wound healing disturbances (REVISION group) on day 16 (mean: 15.6 ± 5.6 days; range: 9–26). Therefore, a large range of clinical parameters (i.e., wound status, sternal stability, hemoglobin, leucocytes, C-reactive protein (CRP), mobilization status, etc.) were analyzed on the ninth postoperative day. A multivariate logistic regression analysis for the REVISION and non-REVISION group was performed. The ROC, based on this model, shows a large area under the curve with satisfying test-characterization by high sensitivity (92.3%) and specificity (89.1%). Of interest, only sternal status (stable vs. nonstable, p < 0.001) and hemoglobin level (11.0 ± 1.3 g/dL vs. 10.1 ± 1.1 g/dL, p < 0.005 for non-REVISION and REVISION group, respectively) were highly indicative for prediction. Leucocyte level was only in trend and CRP was without influence to predict sternal revision. Additionally, there was a high negative predictive value (99.5%), but only a low positive predictive value (32.4%) for sternal revision prediction.

Conclusion: A higher level of hemoglobin and a normal sternal-stability are the only factors predicting the needlessness of sternal revision on a high positive predictive value, whereas increased levels of leucocytes and CRP have only a weak impact and additional wound status parameters have almost no impact on prediction of sternal revision.