Development Of A Prediction Model For Catheter Related Bladder Discomfort In Surgical Patients During Emergence From General Anesthesia

Objective:This study aims to investigate the risk factors of catheter-related bladder discomfort(CRBD)in surgical patients after general anesthesia surgery in the postanesthesia care unit,and to develop and validate a prediction model based on these risk factors.This prediction model may provide decision-making reference for clinicians to identify high-risk patients and use preventive measures.Methods:Between December 2020 to April 2021,we enrolled a total of 411 patients based on the study’s inclusion and exclusion criteria.The following clinical information was collected: baseline characteristic(gender,age,height,weight,smoking status,alcohol used);Preoperative laboratory urinalysis results(urine occult blood,protein,leukocyte esterase),surgical anesthesia related information(Occasion of catheterization,type of surgery,body position,duration of surgery,intraoperative use of dexmedetomidine,intraoperative use of atropine or Penehyclidine,use of oxycodone before the end of surgery,total urine output,etc.),and information in the anesthesia recovery room(CRBD severity classification,postoperative acute pain score,postoperative muscle relaxant antagonist use).According to the whether the occurrence of catheter-related bladder discomfort,patients were divided into catheterrelated bladder discomfort group(CRBD group)and control group(NonCRBD group).Data analysis was performed by using R 4.1 software and R studio 1.2.5042 software.Prediction model was established based on statistical results and clinical experience.Factors with a P value less than0.05 in the univariate analysis and type of surgery,duration of surgery,intraoperative use of atropine or penehyclidine,and intraoperative use of dexmedetomidine were included into a stepwise multivariate logistic regression with alkaike information criterion for variable selection.Prediction model was build based on the final variables in the multivariate logistic regression.The multicollinearity between the variables in the model was tested by variance inflation factor(VIF).Enhance bootstrap resampling method with 1000 replications was used for internal validation.The model’s performance and clinical utility were assessed using the area under the receptor operating characteristic curve,calibration curves,and decision curve analysis.A nomogram was depicted for the clinical use.Results:(1)The overall incidence of CRBD was 45.5% in the included patients(187/411).The results of univariate analysis showed that there were statistically significant differences between the two groups regarding the factors of gender,smoking status,alcohol use,positive occult blood in the preoperative urine routine,positive leukocyte esterase in the preoperative urine routine,the occasion of catheterization,and use of oxycodone before the end of surgery.(P<0.05)(2)The results of multivariate analysis showed that gender,the occasion of catheterization,type of surgery,intraoperative use of dexmedetomidine,use of oxycodone before the end of surgery,and duration of surgery were predictors of CRBD and had value in predicting CRBD.(3)There is no multicollinearity among final variables(All VIF is less than 5).The area under the ROC curve of the prediction model was 0.780(95 % CI: 0.735-0.825);the calibration curve indicated that the model had a good degree of calibration.After enhanced bootstrap(n=1000)resampling to correct the overoptimistic of the prediction model,the area under the ROC curve was 0.759(95 % CI: 0.757-0.76);the corrected calibration curve showed that the prediction model still has a good degree of calibration,but there is a certain degree of underestimation for patients with low CRBD risk and a certain degree of overestimation for patients with high CRBD risk.Decision curve analysis results indicate that the predictive model has potential clinical value.Conclusion:This study preliminarily developed and validated a predictive model to predict the occurrence of catheter-related bladder discomfort in surgical patients after general anesthesia in the postanesthesia care unit.Six variables,namely,gender,the occasion of catheterization,intraoperative use of dexmedetomidine,use of oxycodone before the end of surgery,type of surgery and duration of surgery,had value in predicting catheter-related bladder discomfort and were used for predictive model establishing.The predictive model exhibits good discriminative ability and accuracy,and has potential value for clinical application.