An Internally Validated Risk Model Of Deep Sternal Wound Infection (DSWI)-A Review Of 5,179 Patients Undergoing Index Median Sternotomy Procedures
John P. Fischer, MD, Marten N. Basta, BA, Jason D. Wink, BS, Jonas A. Nelson, MD, Joseph M. Serletti, MD, Stephen J. Kovach, MD, David W. Low, MD, Prashanth Vallabhajosyula, MD, Michael A. Acker, MD, Suhail Kanchwala, MD.
Hospital of the Univesity of Pennsylvania, Philadelphia, PA, USA.
Background: The risk of deep sternal wound infection (DSWI) following median sternotomy is associated with considerable morbidity and mortality. To date there is no comprehensive risk model capable of predicting risk of DSWI to assist in preoperative risk stratification, patient counseling, and delineation of cost. We aim to review a large single center experience with DSWI treated with muscle flaps to create a predictive model and decision-support tool.
Methods: We performed a review of all median sternotomies performed between 5/1/2007-5/31/2013, identifying patients undergoing coronary artery bypass graft (CABG), valve, or aortic root procedures. Preoperative patient comorbidities, defined by AHRQ classifications, were identified from the hospital administrative database. Operative characteristics including internal mammary (IMA) harvest, re-operative median sternotomy, and concurrent procedures were identified. The dependent variable of interest was DSWI (defined as requiring treatment with a muscle flap). Multivariate logistic regression analysis with internal bootstrap validation was performed. Risk factors were weighted based on respective ß coefficients and used to create a sternal wound risk assessment tool (SW-RAT).
Results: Overall, 5,179 patients underwent median sternotomy of which 1.9% (n=99%) developed a DSWI requiring flap. The mainstay flaps were pectoralis major turn-over (N=37) or advancement (N=36) and omental flap (N=11). Index operations included valve procedure (N=3,285), CABG (N=1,653), and aortic root procedure (N=141). Patients were on average 66.9±13.6 years of age, 6.3% were obese (BMI>30 kg/m2), 21.8% were smokers, and 9.4% were redo median sternotomy procedures.
Multivariate regression revealed several factors independently associated with DSWI: female gender (OR=1.7, P=0.03), IMA harvest (OR=2.7, P<0.001), recent weight loss (>10 lbs) (OR=3.1, P<0.001), and obesity (OR=15.1, P<0.001). Risk factors were used to construct a decision-support tool called the SW-RAT (0-10) with a risk distribution of 0.3% to 9.2% (Figure 1). The risk tool significantly discriminated risk across groups: minimal (0.3%), moderate (1.8%), and severe (9.2%) (P<0.001) (C-Statistic=0.80). Ultimately, patients experiencing DSWI requiring flap stayed significantly longer in the hospital (28.4 vs. 13.3 days, P<0.001), more often experienced subsequent unplanned readmission (46.5% vs. 6.5%, P<0.001), were more likely to experience 90-day mortality (18.2% vs. 5.0%, P<0.001), and accrued significantly greater healthcare costs ($180,330 vs. $66,256, P<0.001). As such, the risk stratification tool appropriately demonstrated that higher risk patients experienced increased length of primary hospital stay (P<0.001), more readmissions (P<0.001). higher rates of mortality (P<0.001), and accrued greater cost (P<0.001)
Conclusion: This is the first study to date using preoperatively identifiable patient and operative factors to predict risk of DSWI in median sternotomy patients. The analysis successfully creates a risk stratification tool which incorporates patient (obesity, weight loss, female gender) and operative factors (IMA harvest) into a model that accurately delineates risk of DSWI. This model can serve as a decision-support tool for healthcare systems to better assess risk and as a platform to improve outcomes and patient selection so as to practice more cost-efficient care.
Back to Program