Hyperglycemia as a risk factor for postoperative early wound infection after bicondylar tibial plateau fractures: Determining a predictive model based on four methods

Abstract
Objectives
Identify a glucose threshold that would put patients with isolated bicondylar tibial plateau fractures at risk of early wound infection (i.e. 
Design
Retrospective review of medical records.

Setting
Academic American College of Surgeons (ACS) Level 1 trauma center.

Patients
Adult patients between 2010 and 2015 with an operatively treated isolated bicondylar tibial plateau fracture and at least three glucose measurements during their hospitalization.

Main Outcome Measurement
To predict infection using four different methods: maximum preoperative blood glucose (PBG), maximum blood glucose (MGB), Hyperglycemic Index (HGI), and Time-Weighted Average Glucose (TWAG).

Results
126/381 patients met our inclusion criteria. Fifteen (12%) patients had an open fracture and 30/126 (23%) developed an infection. Median glucose for each predictive method studied was 114 (IQR 101.2–137.8) mg/dL for PBG, 144 (IQR 119–169.8) mg/dL for MBG, 0.8 (IQR 0.20–1.60) mmol/L for HGI, and 120.4 (IQR 106.0–135.6) mg/dL for TWAG. As expected, infected patients had higher PBG, MGB, and TWAG. HGI was similar in both groups. None of these differences prove to be statistically significant (p > .05). Logistic regression models for all the methods showed that having an open fracture was the strongest predictor of infection.

Conclusion
It is well known that stress-induced hyperglycemia increases the risk of infection, we present and compare four models that have been used in other medical fields. In our study, none of the methods presented identified a glucose threshold that would increase the risk of infection in patients with bicondylar tibial plateau fractures.

Level of Evidence: Retrospective review, Level III. See Instructions for Authors for a complete description of levels of evidence.