Description

CR-POPF is a multi-center observational prospective cohort study. The research team focuses on whether perioperative abnormal hyperglycemia can independently predict the occurrence of CR-POPF. It is proposed to realize the dynamic monitoring of perioperative blood glucose level by non-invasive continuous glucose monitoring (CGM), and perioperative continuous glucose monitoring can reflect the overall blood glucose trend and mode, which is helpful to further explore the correlation between perioperative abnormal blood glucose level and CR-POPF.

The investigators expect to enroll 600 patients. Based on our previous clinical experience, the incidence of postoperative pancreatic fistula is approximately 20%. We planned to enroll 600 patients during the 6-month enrollment period in three participating hospitals. The sample size would have a statistical power of 81.3% to detect an odds ratio (OR) greater than 1.34 or less than 0.75.

Measurement data consistent with normal distribution were expressed as mean ± standard deviation, and t-test was used for comparison between groups. When not consistent with normal distribution, the measurement data were expressed as quartiles, and the Wilcoxon-Mann-Whitney rank sum test was used for comparison between groups. Count data were expressed as frequency and percentage. Pearson chi-square test or Fisher's exact test was used for comparison between groups. The binary logistic regression method was used to screen the risk factors and stratify the risk levels, P < 0.05 was considered statistically significant. A generalized estimation equation model was constructed to analyze the correlation between continuous blood glucose monitoring values and postoperative pancreatic fistula. The dependent variable in the model is whether there is pancreatic fistula after operation. The dependent variable in the model was the presence or absence of postoperative pancreatic fistula, and the main variable was the continuous glucose monitoring (CGM) value. The connection function uses the binomial distribution logit function. The multiple measurements of the same patient are set as the group-level variable, and the autocorrelation matrix is selected according to the Quasi-likelihood under Independence Model Criterion (QIC) value of the model. In addition to analysis using generalized estimating equations, continuous glucose monitoring data will be constructed, such as an analysis of the association between pancreatic fistula and the area under the curve for measurement and time beyond normal glucose values. The statistical software was STATA ( version 15.0; stata Corp., Texas, USA ) and R 3.6.1 software ( R Foundation for Statistical Computing, Vienna, Austria ), bilateral P < 0.05 was considered statistically significant.