住院結腸鏡檢查病人腸道準備失敗風險列線圖預測模型的構建與驗證

Construction and validation of a nomogram prediction model for the risk of bowel preparation failure in inpatients colonoscopy examination

LIU Jianying，JIN Caiyun，LAI Lingcui，LU XuefengZhongshan Hospital（Xiamen），Fudan University，Fujian 361000 China

Corresponding Author LU Xuefeng，E-mail：lu.xuefeng@zsxmhospital.com

Abstract Objective：To construct a risk prediction model for bowel preparation failure in inpatients colonoscopy examination and verify its prediction effect.Methods：A total of 491 inpatients who underwent colonoscopy or colonoscopy surgery in a tertiary hospital in Xiamen from February to August 2023 were selected as the survey subjects.All samples were randomly divided into the modeling group （343 cases） and the validation group （148 cases） in a ratio of 7：3.Lasso regression was used to screen risk predictors.Logistic regression analysis was used to identify risk factors for intestinal failure and establish a nomogram prediction model.The prediction effect of the model was evaluated through Receiver Operating Characteristic （ROC） curve analysis and F1 score，and the Hosmer-Lemeshow test was used to evaluate the model′s fit.Results：Logistic regression analysis showed that calcium channel blockers，morning examination，constipation，first colonoscopy，age ≥75 years，and occasional walking were independent risk factors for bowel preparation failure.Including the above variables into the model，the area under the curve （AUC） of the modeling group was 0.725（95%CI 0.664-0.787，Plt; 0.001），the sensitivity was 0.637，the specificity was 0.722，the accuracy rate was 74.64%，the F1 score was 0.843，and Hosmer-Lemeshow test showed that χ2=8.81，P=0.267.The AUC of the validation group was 0.737（95%CI 0.644- 0.830，Plt; 0.001），the sensitivity was 0.795，the specificity was 0.624，the accuracy rate was 77.03%，and the F1 score was 0.856，and Hosmer- Lemeshow test showed that χ2=6.25，P=0.400.Conclusions：The constructed risk nomogram prediction model for bowel preparation failure can provide a reference for the early identification of patients at high risk of bowel preparation failure and the development of targeted bowel preparation intensive programs.

Keywords inpatients;colonoscopy;bowel preparation;predictive model;risk factors;nursing

摘要 目的：構建住院結腸鏡檢查病人腸道準備失敗風險預測模型，并對其預測效果進行驗證。方法：選擇2023年2月—8月在廈門市某三級醫院行結腸鏡檢查或腸鏡下手術的491例住院病人作為調查對象，按照7∶3的比例將所有樣本隨機分為建模組（343例）與驗證組（148例）。使用Lasso回歸篩選風險預測因子，并使用Logistic回歸分析確定腸道準備失敗的危險因素，建立列線圖預測模型。通過受試者工作特征曲線（ROC）分析及F1分數對模型的預測效果進行評價，并使用Hosmer-Lemeshow檢驗對模型進行擬合度評價。結果：Logistic回歸分析顯示，使用鈣通道阻滯劑、上午檢查、便秘、首次行結腸鏡檢查、年齡≥75歲、偶爾行走是腸道準備失敗的獨立危險因素。將上述變量納入模型，建模組ROC曲線下面積（AUC）為0.725［95%CI（0.664，0.787），Plt;0.001］，靈敏度為0.637，特異度為0.722，正確率為74.64%，F1分數為0.843，Hosmer-Lemeshow檢驗示，χ2=8.81，P=0.267。驗證組AUC為0.737［95%CI（0.644，0.830），Plt;0.001］，靈敏為0.795，特異度為0.624，正確率為77.03%，F1分數為0.856，Hosmer-Lemeshow檢驗示，χ2=6.25，P=0.400。結論：構建的腸道準備失敗風險列線圖預測模型可為早期識別腸道準備失敗高風險病人、針對性制定腸道準備強化方案提供參考。

關鍵詞 住院病人；結腸鏡；腸道準備；預測模型；危險因素；護理

doi：10.12102/j.issn.2095-8668.2024.13.012

隨著診斷性結腸鏡檢查技術與內鏡下息肉切除術的發展，結腸鏡檢查已經成為診斷及預防結直腸癌的首選方法，并為直接切除癌前病變和早期癌癥提供了可能［1-2］。為了確保檢測的準確，完整的黏膜檢查和充分的腸道準備必不可少，然而并非所有病人的腸道準備質量能夠達到臨床要求。研究顯示，10%～25%的結腸鏡檢查病人存在腸道清潔不良的情況［3-5］。……