Risk Factors And Predictive Model For Adverse Events Of Painless Gastrointestinal Endoscopy

ObjectivesGastrointestinal (GI) endoscopy is a common technique in diagnosing diseases ofdiagestive system. With the development of anesthesiology, more and more patients tendto choose anesthesia or sedation for GI endoscopy, which is also called painless GIendoscopy. More than80%cases of GI endoscopy were performed under sedation.Propofol, the most popular medication used in sedation for GI endoscopy, is consideredto be safe, but there are still reports about complications during GI endoscopy underpropofol sedation, especially in patients with basic cardiac and pulmonary diseases.Therefore, it is important to understand the indication of sedation and endoscopy. Forthose patients at high risk, more attention should be paid on the monitoring or thesedation should be cancelled. However, there is still no a consensus on the indication ofsedation for GI endoscopy. Many studies showed that grading system proposed by theAmerican Society of Anesthesiologists was not directly correlated with the incidence ofcomplications. Thus, the present study aimed to screen the risk factors and establish apredictive model for adverse event of painless GI endoscopy, so as to provide theoreticalevidence for preoperative screening of patients at high risk.Methods1. Parameters for risk estimation of adverse event during painless GI endoscopyA qualitative research was performed by interviewing specialists from Center of GIendoscopy and Department of Anesthesiology in a tertiary hospitals. The specialistsshould be with a title of associate professor or professor, and they should have a workingexperience of painless GI endoscopy for more than5years. Interview outline wasdesigned on the basis of literatures and anthoritative data, and interview was performedin a semi-structural method. The information was noted by sound record or taking notes,and analyzed by the QST Nvivo8.0software. The adverse events and related factorswould firstly be proposed by the interviews. Then Delphi expert consultation was performed to further establish the assessment system of risk screening. Expertconsultation tables were designed based on the qualitative research. Experiencedphysicians, nurses and anesthesiologists were consulted and they should be with a title ofassociate professor or professor, and with a working experience on GI endoscopy formore than10years. The parameters and weights of adverse events and related risks werefinally determined by two rounds of consultation.2. Risk factors and predictive models for adverse events of painless GI endoscopyA total of844patients undergoing GI endoscopy under propofol sedation wereincluded in the Center of Digestive Endoscopy, Changhai Hospital, the Second MilitaryMedical University, from November to December,2012. Endoscopy was performed afterapproval of ethic committee and informed consent. Simulated electrocardiogram, SpO2and non-invasive blood pressure were monitored during the procedures. Adverse eventsand related factors were recorded. The potential risk factors were further screened byunivariate and multivariate analyses. A mathematical model was established based on theLogistic regression analysis.3. Confirmation of the predictive model for adverse events during painless GIendoscopyA total of600patients undergoing GI endoscopy under propofol sedation wereincluded in the Center of Digestive Endoscopy, Changhai Hospital, the Second MilitaryMedical University, from January to Febrary,2013. Painless GI endoscopy wasperformed similarly to the previous trial. ROC analysis was performed to confirm theefficacy of the predictive models established in the previous trial. The predictedincidence of adverse events was compared to the real one, so as to confirm the accuracyof the models.Results1. Parameters for risk estimation of adverse event during painless GI endoscopyEleven specialists were interviewed by telephone or face-to-face from January toFebrary,2012, including5physicians from Center of Digestive Endoscopy and6 anesthesiologists. The interviews revealed5adverse events, including hypoxia,fluctuation of heart rate and blood pressure, arrythmia, GI perforation and bleeding. Therelated factors contained6tree nodes and18sub-nodes. The tree nodes included generaldata, cardiovascular diseases, pulmonary disease, abnormal airway, other diseases andhistory of medication. The experts consultation involved12experts, including5physicians from Department of Gastroenterology and7anesthesiologists, with a workingexperience of19.25±5.96years. The first round of consultation made severaldefinations of parameters. Hypoxia was defined as a SpO2lower than95%. Flucturationof heart rate and blood pressure was defined as heart rate and blood pressure increased ordecreased by20%of basic values. Old age was defined as an age higher than65years.Overweight or obesity was defined as a BMI higher than25kg/m2. No obvious changewas made by the second round of consultation, other than some modifications on thedescription of parameters. For example, respiratory disease was changed into pulmonarydiseases. After the2rounds of consultation, the final adverse events included5events.The risk factors included6first-class indexes and13second-class indexes.2. Risk factors and predictive models for adverse events of painless GI endoscopyThe844patients included455patients undergoing gastric endoscopy and389patients undergoing enteroscopy. The adverse events included316cases,38cases ofhypoxia,152cases with flucturation of heart rate,105cases with flucturation of bloodpressure,17cases with arrythmia,1case with GI perforation and1case with GI bleeding.The GI complication was not further analyzed because of the low incidence. Theunivariate analysis showed that respiration related factors included higher age old age,overweight or obesity, smoking history, history of cardiac disease and respiratory disease,snoring, and the circulation related factors included old age, overweight or obesity,smoking history, drinking status, history of hypertension, cardiac disease and respiratorydisease. The multivariate analysis was performed to exclude confounding factors. Theresult demonstrated that respiration related factors included old age, smoking history,history of pulmonary disease and snoring. The circulation related factors included old age, history of hypertension, cardiac disease and pulmonary disease. The mathematicalmodels were as follows.Model for respiratory eventsmodel for circulatory events3. Confirmation of the predictive model for adverse events during painless GIendoscopyThe600patients included300patients undergoing gastric endoscopy and300patients undergoing enteroscopy. The adverse events included33cases with hypoxia,127cases with flucturation of heart rate,76cases with flucturation of blood pressure, and12cases with arrhythmia. The ROC analyses showed that the area under the curve was0.699for mathematical model in predicting respiratory events, and0.684formathematical model in predicting circulatory events.The actual incidence of adverseevents was within the95%confidential intervals predicted by the models.ConclusionsThe present study screened the risk factors for respiratory and circulatory adverseevents during GI endoscopy by both the qualitative and quantitative methods. Therespiration-related factors included old age, smoking history, history of pulmonarydisease and snoring; the circulation-related factors included old age, history ofhypertension, cardiac disease and pulmonary disease. The incidence of adverse eventscould be estimated by the number of risk factors, so that patients undergoing painless GIendoscopy at high risk could be screened to ensure the safety of painless GI endoscopy.