Abstract
Introduction
Peri-operative hyperglycemia has many etiologies, including medication, impaired glucose tolerance, uncontrolled diabetes mellitus (DM), or stress, the latter of which is common in post-surgical patients. Our study aims were to investigate if post-operative day 1 (POD1) blood glucose level was associated with post-operative complications after total joint arthroplasty (TJA), including periprosthetic joint infection (PJI), and to determine a threshold for glycemic control that surgeons should strive for during a patient's hospital stay.
Methods
A single-institution retrospective review was conducted on 24,857 primary TJAs performed from 2001–2015. Of these, 13,198 had a minimum one-year follow-up (mean 5.9 years). Demographics, Elixhauser comorbidities, laboratory values, complications and readmissions were collected. POD1 morning blood glucose levels were utilised and correlated with PJI, as defined by the Musculoskeletal Infection Society criteria. The mean age and body mass index of the population was 63.4 years and 30.2 kg/m2, respectively; the sample was comprised of 56.6% females and 48.4% knees. Multivariate analysis was used to determine the influence of several important covariates on complication rate. Youden's J statistic was utilised to determine an optimal blood glucose threshold. An alpha level of 0.05 was used to determine statistical significance.
Results
The rate of PJI significantly increased linearly from blood glucose levels of 115 mg/dL and higher. When examining only patients with a minimum of one-year follow-up, the logistic regression demonstrated that blood glucose levels (odds ratio 1.004 per increment, 95% confidence interval: 1.002–1.006, p=0.001) were a significant risk factor for development of PJI by 1.004 per 1 mg/dL increment. For instance, a patient that had a post-operative glucose level of 280 mg/dL compared to a patient with a blood glucose level of 100 mg/dL had a 2.05 increased odds of developing PJI. The Youden index was used to determine an optimal blood glucose threshold of 137 mg/dL to reduce the likelihood of PJI at one year.
Multivariate analysis revealed that blood glucose levels (OR 1.004, p=0.001), male gender (OR 1.480, p=0.001), body mass index (OR 1.049, p<0.001), operative time (OR 1.004, p=0.006), length of stay (OR 1.059, p<0.001), post-operative hematocrit (OR 0.751, p=0.003), peripheral vascular disease (OR 1.942, p=0.037), liver disease (OR 2.576, p=0.004), rheumatic disease (OR 1.991, p=0.003), and alcohol abuse (OR 2.588, p=0.030) were associated with PJI. Operative or hospital factors that were associated with PJI included the following: low post-operative hemoglobin (OR 0.766 per increment, 95% CI: 0.599–0.979, p=0.034) and longer length of stay (OR 1.145 per day, 95% CI: 1.104–1.187, p<0.001).
The PJI rate in the entire cohort was 1.59% (1.46% in non-diabetics vs. 2.39% in diabetics, p=0.001). There was no association between blood glucose level and PJI (OR 1.002, 95% CI: 0.998–1.006, p=0.331) in diabetic patients, although there was a linear trend for post-operative blood glucose predicting PJI.
Discussion
The relationship between POD1 blood glucose levels and PJI increased linearly, with an optimal cut off of 137 mg/dL. Immediate and strict post-operative glycemic control may be critical in reducing post-operative complications, and even mild hyperglycemia was significantly associated with PJI.
