Data used in this study were obtained from the database of Surgical Site Infection in Orthopaedic Surgery (SSIOS), in which prospective method was used to collect data on patients who underwent orthopaedic surgeries between October 1 and December 31 2018, and surveillance of surgical site during hospitalization and telephone follow-up after discharge were conducted to identify surgical site infections. The ethics committee of the 3rd Hospital of Hebei Medical University approved the SSIOS (NO 2014-015-1), and all the participants had written the informed consent.
Inclusion and exclusion criteria
Patients meeting the following criteria were included: age of 18 years or older, definite diagnosis of tibial plateau fracture, and complete data available. Pathological (metastatic) fracture, old fracture (> 3 weeks from injury), concurrent fractures in other locations, patients with history of DVT or other thrombotic events, or current use of anticoagulants due to chronic comorbidities were excluded from this study.
According our policy, all patients received basic thromboprophylaxis after admission, consisting of chemical (low molecular weight heparin (LMWH), 2500-4100 IU once daily, subcutaneous injection) and elevation of the injured lower extremity for each patient.
Diagnosis of DVT
Guideline for the Diagnosis and Treatment of Deep Vein Thrombosis (3rd edition) proposed by Chinese Medical Association [16] was used to diagnose DVT. Before the operation, routine duplex ultrasonography (DUS) scanning of bilateral lower extremities was performed to detect potential DVT in femoral common vein, superficial and deep femoral vein, popliteal vein, posterior and anterior tibial vein and peroneal vein. The positive criteria of DUS scanning were set as noncompressibility, lumen obstruction or filling defect, lack of respiratory variation in above knee segments, and inadequate flow augmentation to calf and foot compression maneuvers [17]. Due to the less clinical significance, superficial or intermuscular vein thrombosis (soleal or gastrocnemius vein thrombosis) were not included [18, 19].
Data collection and definition
Biomarkers or biomarker-derived inflammatory/immune indexes were obtained from hematologic tests carried out after admission and before the definite operation. These data included neutrophil-, lymphocyte-, monocyte- and platelet counts, The NLR was defined as the neutrophil count divided by lymphocyte count, PLR as the platelet count divided by the lymphocyte count, and MLR as monocyte count divided by lymphocyte count. The systemic immune-inflammation index (SII) was calculated as: platelet count * neutrophil count/lymphocyte count [20]. Given the importance in predictive ability or diagnosis of DVT, palsma D-dimer level was also included.
The other potential factors included demographics (age, gender, body mass index (BMI)), current cigarette and alcohol consumption, the comorbidities (hypertension, diabetes, chronic heart disease), fracture-related factors (injury mechanism (low- or high-energy trauma), open or cloded fracture, fracture classification based on Schatzker classification system).
The BMI (kg/m2) was divided using the criteria recommended by the Chinese working group on obesity: normal (18.5-23.9), underweight (<18.5), overweight (24.0-27.9) and obesity (>=28.0) [21]. Low-energy injury was defined as an injury caused by a fall from a standing height, while fall from a height more than 2 meter or motor accidents was degined as high-energy injury.
Statistical analysis
Continuous variables were expressed by mean and standard deviation (SD), and evaluated by student-t test or Mann Whitney-U test, as appropriate. The categorical data were expressed as number and percentage (%), and were evaluated by chi-square or Fisher's exact test, as appropriate.
For biomarker (neutrophil-, lymphocyte-, monocyte- and platelet counts) and biomarker-derived inflammation/immune indexes (NLR, PLR, MLR and SII) and the plasma D-dimer level in continuous variable, we constructed receiver operating characteristic (ROC) to determine the optimal cut-off value for each variable, when Youden index (sensitivity+specificity-1) was maximum. The significance of the ROC curve was tested using the area under the curve (AUC) analysis, with p<0.05 as significance level. On basis of the cut-off values determined, each variable was divided in to two groups, and the chi-square or Fisher's exact test was performed, as appropriate. we also constructed ROC curve and used the generated AUC to evaluate the discriminatory ability of each biomarker or inflammation/immune index, when they were in dichotomous variable.
In the multivariate logistics regression model, the included variables were those tested as statistially significant in the univariate analyses. The stepwise backward elimination method was used to exclude variables not significantly affecting the development of DVT. In the final model, variables with p<0.10 were retained, and the correlation strength is indicated by odd ratio (OR) and 95% confidence interval (95%CI). The significance level was p<0.05. Fitting degree of the final model was evaluated by Hosmer-lemeshow (H-L) test, and p>0.05 indicated the acceptable result. SPSS23.0 was used to perform all the tests (IBM, armonk, New York, USA).