Sternal dehiscence after sternotomy for cardiac operations severely affect the patients’ physical and psychological recovery. It is also a matter of economics, as complicated postoperative courses are up to 2.8 times more costly than uncomplicated cases. The complication of sternal instability alone is discussed only in some studies and reported to occur in 3% (0.5 – 8.0%) of patients after sternotomy [11,13]. Preventive strategies comprise the usage of vacuum-assisted wound closure therapy, special surgical techniques to reinforce the osteosynthesis and the development of models to identify patients at high risk for sternal wound complications.
Several models focus on special surgical settings like the usage of bilateral internal thoracic artery grafting in coronary artery bypass grafting [10]. Moreover, some models were created to prevent infections on multiple surgical sites (sternum, legs) [3,5], the development of infections alone [3,5,6] or in combination with sternal instability [14]. Although sternal wound complications have a complex pathogenesis depending on specific comorbidities as well as pre-, intra– and postoperative factors, we consider the sternal instability as the initial step towards further sternal wound healing complications including deep sternal wound infections and mediastinitis. In this context, the development of a model to predict the risk of sternal dehiscence may be useful to calculate patients’ risk rationally and individually and to target these patients in the operating room for special surgical intervention strategies.
The aim of this study was both to analyse the risk factors for sternal instability and develop a predictive risk score based on the results of this analysis. Surgeons who operate their patients via sternotomy would be able to evaluate the risk of sternum instability individually, and may optimize the sternal osteosynthesis technique. Up to now, there is no scoring system specifically created to predict the individual risk for sternal instability due to mal-union of the sternum (MUST).
Among all patients (n=8.615) who underwent all-type cardiothoracic operations in a quaternary university hospital setting over a ten-years period from 2008 to 2017 mal-union of the sternum (MUST) occurred in 2.5% of the patients (n=176).
Other authors reported a higher sternal complication rate for wire cerclage usage compared to rigid plate fixation of the sternum (5% vs. 0% after 6 months) [4]. Although rigid plate fixation was associated with a trend toward greater index hospitalization costs, 6-month follow-up costs tended to be lower. As a result, total costs were similar between groups.
Ramann et al. [14] investigated the sternal bone healing in a multicentre RCT and compared pain scores and narcotic usage in patients who received rigid plate fixation versus conventional wire cerclage. He could find a sternal union rate of 70% at 6 months in patients with rigid plate fixation compared to 24% in patients with sternal wiring. Pain scores and narcotic usages were significantly lower in the plate fixation group.
Predictive scoring systems or computed scales for the occurrence of surgical complications
Those predictive scales are instruments to optimize patients’ outcome. Patients at a high risk for a complicated course may need more attention than patients at standard risk.
In 1992, O’Connor et al. [7] developed a multivariate clinical prediction rule using logistic regression analysis, a statistical method that allows the calculation of the conditional probability of death (in-hospital mortality associated with CABG surgery). The area under the ROC curve obtained from the training set of data was 0.74 (perfect, 1.0). The prediction rule performed well when used on a test set of data (area, 0.76). The correlation between observed and expected numbers of death was 0.99.
Prediction of major infections after cardiac surgery
Other studies focused on major all- site infections after cardiac surgery. As published in 2005, Fowler et al. [3] analysed 331.429 patients (operated 2002-2003 for CABG) from the STS National Cardiac Database to identify risk factors for major infection. A simplified risk scoring system of twelve variables accurately predicted risk for major infection (overall, 3.5%; of them, 25% mediastinitis, 33% saphenous harvest site infection, 35% septicaemia, 7% multiple sites). Patients with a major infection had a significantly higher mortality than patients without an infection (17.3% versus 3.0%).
Hussey et al. [5] performed a more specific analysis of sternal wound infections after cardiac surgery in 1998. In this study, a sternal wound infection prediction scale (SWIPS) was developed and further revised (SWIPS-R). A multivariate logistic regression with 12 risk factors provided up to 76% correct predictions.
The scoring systems were mainly considered for sternal wound infections generally after any cardiac operation [3, 5] or more specially like CABG surgery with one or bilateral use of the BITA, analysed by Gatti et al. [10]. With this special scoring system inaugurated by Gatti, deep sternal wound infection (DSWI) after BITA grafting, can be predicted with a accuracy of 0.72 to 0.73 (AUC of the ROC-curve).
Until now, the most effective prevention methods including scoring systems have not been found. They remain objects of an ongoing discussion.
In a review of 2002, Losanoff, Richmann and Jones [17], described not only preoperatively known risk factors, such as insulin-dependent diabetes mellitus, chronic obstructive pulmonary disease, body- mass- index above 35 kg/m² (BMI >35 kg/m²), obesity in diabetic women. He also described intraoperative factors like prolonged bypass time, sternal devascularisation (by internal thoracic artery use) and harvesting technique (whether skeletonized or pedicled technique). Additionally, the authors pointed out that suboptimal primary sternal closure should be considered a significant intraoperative risk factor. This explains the necessity of a predictive scoring system that may help the surgeon in decision –making concerning the most appropriate sternal closure technique. This is called ‘a tailored approach’ in a publication of Nenna et al. [18]. They proposed a decisional algorithm based on clinical experience. The definition of high-risk patients comprises COPD, obesity, BITA use, diabetes, off-midline sternotomy. An algorithm for tailored sternal closure techniques is based on patients’ risk factors and the surgeons’ experience. The authors of this review are aware of the absence of consensus in the literature and the severe lack of RCT regarding the optimal sternal closure method.
Our scoring model is based on a multivariable logistic regression model, it predicts a sternal dehiscence more precisely than other scoring systems (AUC: 0.76). The risk factors are individually weighted and therefore, a more specific surgical technique can rationally be chosen and eventually higher costs are well justified.
How to interpret the predictive scoring system (MUST-Score)
In the first step, the patients diagnoses are weighted with points in a scale from -1 to 5 (Table 4). Secondly, the total score represents the expected risk of sternal dehiscence within a range from 0.2% to 73.7% (Table 5).
The scaling is arbitrarily divided into a coloured traffic light scheme starting from
- green (score up to 4 points, representing an expected low risk of MUST below 1%), to
- yellow (score up to 8 points, intermediate risk of MUST below 5%), to
- red (score up to 11 points, high risk of MUST above 15%) (Table 6).
Table 6: Classification of risk groups and surgical technique modification
Score
|
Risk
|
|
Surgical technique
|
0
|
0.3%
|
low
|
Standard:
|
1
|
0.5%
|
|
Single wires
|
2
|
0.7%
|
|
|
3
|
1.0%
|
<1.0%
|
|
4
|
1.4%
|
Intermediate
|
Standard plus:
|
5
|
2.1%
|
|
Special wiring techniques,
|
6
|
3.0%
|
|
e.g. figure of eight,
|
7
|
4.3%
|
<5%
|
intercostal, double wires
|
8
|
6.1%
|
high
|
Reinforced:
|
9
|
8.7%
|
|
Wires plus bands or
|
10
|
12.2%
|
<15%
|
rigid plates fixation*
|
11
|
16.8%
|
Very high
|
Specially reinforced:
|
12
|
22.7%
|
|
Rigid plates fixation (360-plates +bands**)
|
13
|
29.9%
|
<30%
|
(minimum 3 plates, 5 bridges)
|
14
|
38.3%
|
Extremely high
|
Combination of plates and bands (360°)
|
15
|
47.5%
|
<50%
|
other techniques:
|
16
|
56.9%
|
>50%
|
Additional to Plates
|
17
|
65.8%
|
|
Retention sutures
|
>18
|
73.7%
|
|
Vacuum-assisted wound closure
|
Legend:
*SternaLockTM Blue, **SternaLockTM 360
The expected risk of malunion increases with the total score. As shown in figure 2, the graph follows an exponential curve with an increase behind a score of 10. We consider the risk as high and recommend specially reinforced techniques for the sternal closure as described in table 6.
The surgeon may preoperatively calculate the patients risk group and plan the modification of the sternal closure technique. We consider this a well-tailored approach with an underlying rational score that fits the patients’ needs as well as economical requirements.
Study Limitations
Even the best scoring system does not relieve the surgeon from correct judgement and proper technical quality of the employed technique. Several clinical circumstances may limit the guidance of scores and the usage of extended techniques, e.g. diffuse bleeding with a certain probability of postoperative reexploration and hemodynamic instability.
From the statistician’s point of view, the probabilities have to be interpreted with caution. The analysed complication (sternal dehiscence) is a rare event (2.5%), thus the logistic regression model is lacking of some accuracy in the marginal zones (nearby 0% and 100%). Exclusion criterion was the primary use of any other complex techniques for sternal osteosynthesis, so that some of the so-called high-risk patients potentially did not contribute to the complication group. As well as data of juvenile patients below an age of 18 years were not enrolled.
Experience and availability of special techniques cannot be taken for granted in each centre.
Further studies will have to validate the precision of the performance of this scoring system. In this future validation study, all data would be analysed that have been entered in the database after inauguration of the MUST-score in this department.