Is postoperative glucose variability associated with adverse outcomes following shoulder arthroplasty?

Background: Postoperative infection after shoulder arthroplasty is a devastating complication. Multiple patient risk factors have been associated with postoperative infection, including increased body mass index and diabetes. Although the association between preoperative glucose control and infection has been established, little is known about the effect of perioperative glycemic control on outcomes following shoulder arthroplasty. The purpose of this study was to investigate the association between postoperative glycemic variability and short-term complications after total shoulder arthroplasty (TSA) and reverse total shoulder arthroplasty (RTSA). Methods: A retrospective study was performed at a large, single-specialty center from January 2015 to December 2018. Patients were included if they underwent primary TSA or RTSA with a minimum of 90 days' follow-up and had a minimum of 1 serum glucose value obtained per day of the hospital stay or had ≥3 measurements obtained during the hospital admission period. The primary outcome variable was postoperative infection per accepted definitions of surgical-site infection or periprosthetic joint infection. Secondary outcome variables included stiffness, periprosthetic fracture, periprosthetic dislocation, and reoperation. Results: In total, 1074 TSAs or RTSAs (in 1032 patients) met the eligibility criteria. The mean patient age was 69.9 ± 8.4 years, and 40.3% of patients had a preoperative diagnosis of diabetes mellitus. Of the patients, 670 (62%) had a calculable coefficient of variation. A younger patient age (median, 65 years [interquartile range (IQR), 13.5 years] vs. 71 years [IQR, 11.0 years]; P =.02) and a preoperative diagnosis of diabetes mellitus (P =.01) showed statistically significant associations with postoperative infection. The first in-hospital glucose measurement beyond the reference tertile of 70-140 mg/dL showed a statistically significant association with postoperative infection, with a median of 128.0 mg/dL (IQR, 43 mg/dL) vs. 167.5 mg/dL (IQR, 37.0 mg/dL; P =.01), whereas the second and third glucose measurements showed no association with postoperative infection. We found no associations between the coefficient of variation and reoperations or complications including surgical-site infection, periprosthetic joint infection, death, postoperative infection, periprosthetic fracture, or stiffness. Conclusion: We found an association between a preoperative diagnosis of diabetes mellitus and postoperative infection following shoulder arthroplasty. We also found that an elevated first glucose measurement is associated with the development of postoperative infection. In-hospital glycemic control, as well as preoperative glycemic control and optimization, may be beneficial for reducing postoperative infections following shoulder arthroplasty.