Diabetes is a common disease in perioperative patients, and nearly half of patients with type 2 diabetes mellitus (T2DM) require surgery in their lifetime (1). Studies have shown that adverse events such as surgical site infection, secondary surgical intervention, and death are more common in diabetic patients than in nondiabetic patients (2) and that good perioperative blood glucose control can reduce the occurrence of these adverse events (3). Glycated hemoglobin (HbA1c), a standard blood glucose control measure, reflects the blood glucose level in the past 2–3 months, but it cannot reflect real-time blood glucose changes and is easily affected by factors such as hemoglobin and pregnancy (4). The pain caused by frequent fingertip blood glucose measurements reduces patient compliance. In recent years, with the rise of continuous glucose monitoring (CGM) instruments, CGM has been widely used in clinical practice and has successfully solved the above problems. The time-in-range (TIR), a derived index of CGM, has been widely used to evaluate the level of glycemic control (5). TIR generally refers to the time or percentage of glucose within the target range (usually 3.9~10.0 mmol/L) within 24 hours.

As a clinical tool that can predict the risk of outcome, prediction models have been mostly mentioned in recent years for postoperative infections in cancer or nondiabetic patients. However, infection, a common postoperative complication in diabetic patients, has received little attention. During clinical practice, we found that T2DM patients with higher TIR levels had fewer postoperative infections. For further verification, we designed this study to explore the relationship between preoperative TIR levels and postoperative infections in patients with perioperative T2DM. Moreover, whether the TIR can be used as a new indicator to predict the risk of postoperative infection in patients with T2DM is unknown.

We included 806 patients with type 2 diabetes in the perioperative period from September 1, 2022, to July 31, 2024, at the First Affiliated Hospital of Bengbu Medical University, all of whom met the criteria for the diagnosis and classification of Guideline for the Prevention and Treatment of Type 2 Diabetes Mellitus in China (2020 edition) (6). After admission, the patients uniformly donned a CGM (product model: FreeStyle Libre; manufacturer: Abbott Diabetes Care UK) to monitor blood glucose, and a professional in-hospital blood glucose management team conducted blood glucose control on the basis of blood glucose conditions. During the observation period, there were 23 cases of abnormal damage and detachment of CGM probes, 68 cases of inflammation indicators were not recorded after surgery, 31 cases were transferred to the intensive care unit (ICU) ward after surgery, 22 cases changed the treatment plan without surgery, and 6 cases died. The final number of enrolled patients was 656. According to International Consensus on Use of Continuous Glucose Monitoring (7) and the International Consensus on Time in Range (8), the TIR of adults with type 1 or type 2 diabetes individuals should be greater than 70%. And Richard et al. have previously confirmed the recommendation of using TIR 70% as the baseline target in their study on the risk of adverse events associated with TIR in T2DM patients, and when TIR>70%, severe hypoglycemia and microvascular events are lower (9). Therefore, we chose TIR 70% as the cutoff point, and divided enrolled patients into two groups, TIR≥70% and TIR<70% as exposure factors, and postoperative infection during hospitalization as the research outcome. In addition, all patients were divided into training sets and validation sets at a ratio of 7:3 for analysis of the prediction model ( Figure 1 ). The general information and relevant laboratory test data of all enrolled patients were collected. This study was approved by the Research Ethics Committee of the First Affiliated Hospital of Bengbu Medical University (approval number: 2022088).

As of 2021, the International Diabetes Federation (IDF) estimates that the global population with diabetes has reached 537 million, accounting for 10.5% of the global population; by 2030, the prevalence of diabetes is expected to increase to 643 million (11.3%) (14). As the diabetic population continues to increase, the number of diabetic patients requiring surgical intervention will also increase. Studies have confirmed (15, 16) that poor blood glucose control is an independent risk factor for postoperative infections in various surgical operations. Postoperative infections often lead to a series of problems, such as poor wound healing, prolonged hospitalization, increased medical expenses, and even increased patient risk of death. The World Health Organization proposed that intensive perioperative hypoglycemia should be implemented in diabetic patients to reduce the risk of postoperative infection. However, patients undergoing strict blood glucose control may experience severe hypoglycemia, and the optimal blood glucose control range is still controversial (17). Therefore, finding a perioperative glucose-lowering target to reduce postoperative infections is crucial. CGM is a continuous blood glucose monitoring technology that changes the blood glucose monitoring method from blood to interstitial fluid and integrates and divides all monitored blood glucose values into TIR, TAR and TBR. TIR has become a popular choice for clinicians and patients because it can more intuitively express blood glucose control levels.

The modified Poisson model is a statistical method that can calculate the relative risk on the basis of robust error variance, which is more accurate than the commonly used logistic regression results. According to the International Consensus on Time in Range (8), we recommended a TIR of 70% as the cutoff point to study perioperative T2DM patients. Through modified Poisson regression analysis, we found that the TIR is a risk factor for postoperative infection in T2DM patients, and when the TIR is <70%, the risk of postoperative infection in T2DM patients increases by 52.2%. Thus, for perioperative T2DM patients, poor preoperative TIR control leads to an increased risk of postoperative infection, and blood sugar should be strictly controlled before surgery.

HbA1c, which is used as a standard to evaluate blood sugar control levels, is often affected by multiple factors and fails to reflect blood sugar fluctuations, which are related to diabetic complications (18, 19). In 2017, Jourdan et al. reported that HbA1c was related to postoperative deep infection in T2DM patients who underwent total hip arthroplasty (TKA), but HbA1c could not be used as an independent risk factor for deep infection after TKA (20). In contrast to HbA1c, the TIR can better reflect recent blood sugar fluctuations in real time, and the TIR is better than HbA1c in terms of the operability of blood sugar monitoring and the prediction of diabetic complications (21). As one of the manifestations of blood glucose fluctuation amplitude, the higher the TIR value, the smaller the blood glucose fluctuation amplitude, indicating that the blood glucose is in a relatively stable state (22). When blood glucose fluctuations are too large, the antioxidant capacity of the cells themselves decreases, which may aggravate cellular oxidative stress, leading to damage to endothelial cell function and the induction of inflammatory reactions (23). Experimental studies have shown that prolonged hyperglycemia or excessive fluctuations in blood sugar can reduce the immune function of cells at the site of infection (24). Liu et al., in their study on postoperative infection after lumbar fusion in patients with T2DM, reported that the amplitude of preoperative blood glucose fluctuations was closely related to postoperative infection and could be used as an independent risk predictor of postoperative infection after this surgery (25). TIR is negatively correlated with the amplitude of blood glucose fluctuations. We can use CGM to monitor blood sugar and increase the percentage of TIR to reduce the amplitude of blood glucose fluctuations (26), thereby reducing the risk of postoperative infection. This is consistent with the fact that strict control of blood sugar can reduce the risk of postoperative infection. This finding is consistent with the view of postinfection risk (27).

Although the TIR is an influencing factor for postoperative infection in T2DM patients, can the TIR be used to predict the risk of postoperative infection in T2DM patients? To this end, we established a clinical prediction model that can identify potential correlations between preoperative indicators and postoperative infection outcomes, allowing clinicians to make more accurate predictions of the risk of postoperative infection in T2DM patients so that relevant measures can be taken in a timely manner to reduce this risk. In this study, we used two algorithms to screen variables, among which Lasso regression stands out in data analysis and collinearity processing through automatic variable selection and sparse modeling, while Boruta algorithm performs more outstandingly in nonlinear relationships, with strong stability and can directly provide feature importance ranking and classification results. To eliminate the collinearity of the model, we used the dual method of LASSO regression and the Boruta algorithm to determine the accuracy of the feature variables and the stability of the prediction model (28). Finally, both LASSO regression and the Boruta algorithm revealed that the TIR was included in the clinical prediction model as a characteristic variable of postoperative infection in T2DM patients. To increase the readability of the model and help clinicians obtain relevant risk probabilities through simple calculations, we presented all characteristic variables in the form of nomograms.

In the clinical prediction model, in addition to TIR, we included NEU, LYM, TC, SOD and surgery type. HbA1c was not included in part because anemia and hypoalbuminemia affect the accuracy of HbA1c in some patients with tumors or pregnant caesarean sections, and the inability of HbA1c to respond to blood sugar fluctuations in a timely manner is also a major drawback. WBC, NEU, LYM, and CRP are commonly used inflammatory markers in clinical practice, and are more easily obtained from routine examinations compared to inflammatory factors such as tumor necrosis factor or interleukin. But studies have shown that CRP has a weak correlation with metabolic diseases and is easily affected by HDL (29). WBCs have poor sensitivity and specificity, and NEU is at the forefront of the body’s defense against pathogenic invasion and can first reach the site of inflammation when infection occurs (30). LYM plays a core role in the body’s immune response and can respond to viral infections in a timely manner. Therefore, jointly incorporating NEU and the LYM into the prediction model is better than incorporating a single WBC indicator. TC is the major steroidal compound in mammals and may trigger inflammatory cascades when regulating intracellular and intracellular homeostasis (31). Morimoto et al. also confirmed in a study on total serum cholesterol and postoperative internal infections of the gastrointestinal tract (32) that there was a reverse-J-type correlation between total serum cholesterol and postoperative internal infections, which may be due to the ability of circulating cholesterol rich lipoproteins and triglyceride rich lipoproteins to bind and detoxify bacterial lipopolysaccharides. Thus, TC was used as a reliable predictor of postoperative infection. SOD is an antioxidant metalloenzyme that exists in organisms. It plays a vital role in oxidative stress in the body and is inseparable from infection. It is associated with early death in acute pancreatitis and stroke infection (33, 34). In a mouse experiment by Christina et al. (35), SOD also showed a protective effect during lung infection. This is consistent with the fact that, in this study, the risk of postoperative infection gradually increased as SOD levels decreased. The results are consistent. The Centers for Disease Control and Prevention classify surgical incisions into 4 categories: Type I/clean incision, Type II/clean-contaminated incision, Type III/contaminated incision, and Type IV/infected incision (36). Studies have shown that the type of surgical incision is related to postoperative infection (10). Our study also revealed that, as the level of incision type increases, the risk of postoperative infection increases, which suggests that we can take corresponding preventive measures according to the different types of postoperative incisions used in T2DM patients.

In this study, we also conducted a series of subgroup analyses, and the results suggested that the TIR is robust among subgroups and is an independent prognostic factor for postoperative infection in patients with T2DM. Notably, this study retained some Class III incisions that met the inclusion criteria. Subgroup analysis revealed that, when the surgical incision type was a Class III incision, even if the TIR was ≥70%, more people would develop postoperative infections. This finding suggests that when the postoperative incision is contaminated, although preoperative blood sugar control is acceptable, the risk of postoperative infection is still high. We should take relevant preventive measures in a timely manner after surgery to reduce the risk of postoperative infection.

In recent years, composite indicators such as TyG, SHR, and THR have been shown to be related to the risk of cardiovascular and cerebrovascular diseases, sepsis, and postoperative gastrointestinal infections (37–39). Therefore, this study also included these new indicators. The composite indicators were analyzed, but the results suggested that these indicators were not related to postoperative infection in patients with T2DM. This may be because these indicators are strongly related to single-disciplinary surgeries, and there were many types of surgeries in our study that were not limited to gastrointestinal surgery, tract surgery or cardiac surgery and therefore were not suitable for this study.

In conclusion, our study shows that the TIR is not only a risk factor for postoperative infection in patients with perioperative T2DM but also has good predictive value for postoperative infection. When the preoperative TIR is lower than 70%, TIR is less likely to cause postoperative infection in T2DM patients. The risk will be greatly increased. In most previous studies, prediction models have been used only to predict diabetic complications or surgical outcomes, and few studies have investigated postoperative infection in patients with T2DM. This model is more helpful for clinicians to evaluate conditions on the basis of individual patient conditions, reduce the occurrence of postoperative infections and shorten the length of hospitalization. However, this study also has shortcomings. This was a single-center study that lacked external verification, only observed the occurrence of infections during hospitalization, and cannot be used to evaluate the situation after discharge. Therefore, the next step is to evaluate the long-term predictive value of the model after surgery, and the accuracy of the model will be further tested through external validation.