Systematic Review And Effectiveness Study Of Pressure Injury Risk Prediction Model For Surgical Patients

Objectives:This study aims to systematically evaluate the risk prediction model of Intraoperatively Acquired Pressure Injury(IAPI)in surgical patients,and provide a reference for the construction and optimization of the IAPI risk prediction model.The IAPI risk prediction model was applied to the IAPI risk assessment of cardiovascular surgery patients to explore its accuracy and effectiveness in clinical practice.At the same time,the application effect of the two types of assessment tools,the risk assessment scale and the risk prediction model,were compared and analyzed in the same surgical population,so as to provide a basis for the selection of IAPI risk assessment tools for patients undergoing cardiovascular surgery.Methods:(1)Systematic review of the IAPI risk prediction model:Major English and Chinese databases were systematically searched and references in the included literature were traced to relevant research on IAPI risk prediction models for surgical patients.The performance and model presentation were summarized and analyzed,and the risk of bias and applicability of the included models were evaluated using the PROBAST.(2)Preliminary application of the IAPI risk prediction model for cardiovascular surgery patients:According to the results of the systematic review,a well-performing,easy-to-operate,and accessible IAPI risk prediction model for cardiovascular surgery patients(Lu CX prediction model)was facilitated and evaluated for cardiovascular surgery patients in two tertiary care hospitals in Lanzhou from January 2022 to March 2023.At the same time,the prediction model was compared with the IAPI risk assessment scale(the Scott Triggers score scale and the Braden scale).Excel and SPSS software were used to establish a database,and Med Calc software was used to draw the Receiver Operating Characteristic(ROC)curve.The area under the ROC curve(AUC),Sensitivity,specificity were calculated to analyses the actual prediction effect of the two types of evaluation tools.Results:(1)Systematic review of IAPI risk prediction models:A total of 21 IAPI risk prediction models were included,including 20 model development studies and 1model validation study.Eighteen models(85.71%)were published in the past five years.Seven models had a target population explicitly aimed at patients undergoing cardiovascular surgery.Most of the models(16 models)were established based on the Logistic regression algorithm.Fourteen models reported AUC values ranging from0.727 to 0.897,five models reported C-index ranging from 0.725 to 0.815,and two models reported C-statistics of 0.78 and 0.984,respectively.Eleven models reported classification ability indicators for the models with sensitivities ranging from 8.11%to98%,specificities ranging from 19%to 100%,positive predictive values ranging from6.0%to 100%,and negative predictive values ranged from 76.71%to 99.1%.Calibration was reported as a calibration plot for 8 models and Hosmer-Lemeshow goodness-of-fit test for 4 models.Twelve models were validated internally.The number of predictors included in the model ranged from 2 to 13,and the top three predictors in terms of frequency of occurrence were operation time,age and body mass index.The final model categories of the included models are mainly nomogram models and regression models.According to the PROBAST evaluation criteria,the applicability of the 21 models was good,but there was a certain risk of bias,with the high risk mainly in the two fields of"study population"(9 models)and"analysis"(21 models).(2)Preliminary application of the IAPI risk prediction model(Lu CX prediction model)in patients undergoing cardiovascular surgery:A total of 221 patients participated in the study,of whom 29 patients developed IAPI,with an incidence rate of 13.12%.The degree of IAPI is stageⅠand stageⅡ,and the site of IAPI is mainly in the sacrococcygeal region,and the area of IAPI ranged from 0.25cm~2 to 50cm~2.The predictive effects of three assessment tools(Lu CX predictive model,ST score scale,Braden scale)were analyzed.The sensitivities of the three assessment tools were68.97%,96.55%,93.10%,and the specificities were 70.83%,72.40%,45.31%,respectively.The ability of the Lu CX predictive model to identify IAPI patients from non-IAPI patients was lower than that of the other two assessment tools.The ST score scale had better ability to detect IAPI patients and non-IAPI patients.The Braden scale had better ability to screen IAPI patients but weaker ability to identify non-IAPI patients.The scores of the Lu CX prediction model and the ST score scale were negatively correlated with the prediction outcome(P<0.05).The AUC of the Lu CX prediction model was 0.712(0.591,0.833)at a cut-off value of-0.29,which had a certain accuracy for the diagnosis of IAPI.The AUC of the ST score scale was 0.785(0.725,0.837)at a cut-off value of 2,which had a moderate predictive value for the diagnosis of IAPI.The AUC of Braden scale was 0.598(0.530,0.663)at a cut-off value of 12,which was less than 0.6 and had a low predictive value for IAPI in patients undergoing cardiovascular surgery.Conclusion:The IAPI risk prediction models constructed so far have good discrimination and impressive overall performance.Most of the models adopt internal validation to examine the performance indexes of the models,and future researchers should actively carry out external validation and update the models to clarify the reproducibility and generalizability of different models.The IAPI risk prediction model(Lu CX prediction model)is relatively simple to operate and has moderate distinguishing ability when used for IAPI risk assessment in patients undergoing cardiovascular surgery.In the future,independent risk factors for the occurrence of IAPI in cardiovascular surgery patients should continue to be explored,and the risk prediction model should be further optimized and improved before being extended to clinical practice.