Abstract :
BACKGROUND CONTEXT: Surgical site infections (SSI) is one of the most common complications in spinal surgery which potentially leads to higher morbidity, mortality, extended length of hospital stays, and increased health care costs. Currently, there is a validated risk stratification model, the Spine Sage model that specifically predicts the likelihood of surgical site infection after spinal surgery requiring return to the operating room for surgical debridement. The value of this model is that it gives the user in percentage the absolute probability of developing a post-operative SSI depending on the surgical invasiveness and patient’s comorbidities. Patients are much more likely to understand an absolute percentage, rather than relative risk and confidence interval values. A model like this is of paramount importance to counsel patients and foresee the safety of a spinal procedure.
PURPOSE: To evaluate the ability and discriminatory power of the "Spine Sage" model to predict the risk of surgical site infection requiring surgical debridement after posterior lumbar fusion.
OUTCOME: See if patients who have had a surgical site infection and have had surgical debridement were predicted by the “Spine Sage “ model and if so, from what percentage is the patient really at very high risk to make an infection of the operative site requiring a return to the operating room.
METHODS: We carried out a retrospective case-control study, by the use of the computerized files patients of the HUG neurosurgery department to collect information concerning the risk of infection of 50 patients operated on between 2011 and 2014 for posterior lumbar fusion because of degenerative lumbar pathologies. Of these 50 patients, N=7(14%) were detected with postoperative surgical site infection (SSI group) and N=43(86%) without postoperative surgical site infection (non- SSI group). The part of Spine Sage model predicting the chance to return to operating room for a SSI debridement was retrospectively applied to the whole cohort. The discriminatory power of Spine Sage model to predict SSI requiring a new operation was then calculated and compared between the two groups. The risk factors required for calculating the infectious risk by the "Spine Sage" model were extracted and were entered into the website www.spinesage.com in order to calculate for each patient the Spine sage value. The data is summarized by group (SSI group vs non-SSI group) using the mean (SD) for continuous variables or median (range) when the normality assumption was violated. For categorical variables, the summary is given by numbers and percentages. Identification of variables associated with the SSI group is performed by univariate logistic regression.
RESULTS: The univariate logistic regression showed that the risk of infection increases linearly with the Spine Sage score. The gain of one point of Spine Sage increases the risk of infection by 25% and this increase is almost significant p value = 0.052. The discriminatory power of the spine sage score including all data (N = 50) measured using the area under ROC curve was of 0.67 which is considered like a not very good discriminating ability of the Spine Sage model to predict SSI.
CONCLUSION: There are currently very few models that predict the risk of SSI requiring a return to the operating room for debridement and the Spine Sage model that we evaluated in our study is not very prone to predicting it and this probably because the small size of our sample and its homogeneity in terms of pathologies.