颅脑创伤后垂体功能减退的列线图预测模型构建_陈艾.pdf

现代医学Modern Medical Journal2023，Apr;51(4):447-454 收稿日期2022-11-23 修回日期2023-03-27 作者简介陈艾(1982 )，男，土家族，贵州沿河人，主治医师。E-mail:nuoh717378163 com 通信作者罗涛E-mail:715707071 qq com 引文格式陈艾，程波，苏俊，等 颅脑创伤后垂体功能减退的列线图预测模型构建 J 现代医学，2023，51(4):447-454 论著 颅脑创伤后垂体功能减退的列线图预测模型构建陈艾，程波，苏俊，罗涛(重庆市南川区人民医院 神经外科，重庆408400)摘要目的:构建颅脑创伤后垂体功能减退的预测模型并验证。方法:选取 2021 年 1 月至 2022 年 5 月我院进行治疗的 620 例颅脑创伤患者为研究对象，按照 7 3 的比例随机分为建模组(434 例)和验证组(186例)，其中建模组根据是否发生垂体功能减退分为减退组和正常组;收集患者临床资料，分别采用单因素和多因素 Logistic 回归分析影响颅脑创伤后垂体功能减退的危险因素;构建预测颅脑创伤后垂体功能减退的列线图模型，采用校正曲线、H-L 拟合优度检验评价列线图模型的校准度，绘制受试者工作特征(OC)曲线验证模型的区分度。结果:建模组434 例颅脑创伤患者有143 例发生垂体功能减退，发生率为32 95%;多因素Logistic 回归分析显示入住 ICU(O=12 644，95%CI 5 800 27 566)、入院 GCS 评分8 分(O=8 168，95%CI 2 478 26 927)、弥漫性脑水肿(O=5 759，95%CI 2 329 14 241)、脑疝(O=2 220，95%CI1.035 4 762)、中线移位5 mm(O=13 479，95%CI 6 640 27 360)、颅内压增高(O=6 957，95%CI2 459 19 682)、颅底骨折(O=2 538，95%CI 1 083 5 950)、住院天数(O=1 136，95%CI 1.079 1.197)均为影响垂体功能减退的独立危险因素(P 0 05)。构建列线图模型并进行内外部验证，H-L拟合优度检验结果显示，建模组的 2=7 287，P=0 506，验证组的 2=7 202，P=0 515，一致性较好;OC 曲线分析结果显示建模组和验证组预测颅脑创伤患者垂体功能减退的 OC 曲线下面积(AUC)分别为 0 929(95%CI 0 906 0 953)、0 892(95%CI 0 843 0 942)。结论:入住 ICU、入院 GCS 评分8 分、弥漫性脑水肿、脑疝、中线移位5 mm、颅内压增高、颅底骨折、住院天数均为影响颅脑创伤后垂体功能减退的危险因素，基于以上危险因素构建的预测模型可有效预测颅脑创伤后垂体功能减退的风险，有助于临床医师早期识别颅脑创伤后垂体功能减退患者。关键词颅脑创伤;垂体功能减退;影响因素;列线图;预测模型 中图分类号651 1 文献标志码A 文章编号1671-7562(2023)04-0447-08doi:10 3969/j issn 1671-7562 2023 04 004Establishment of nomograph model for individualized prediction ofhypopituitarism after craniocerebral traumaCHEN Ai，CHENG Bo，SU Jun，LUO Tao(Department of Neurosurgery，Chongqing Nanchuan District Peoples Hospital，Chongqing 408400，China)Abstract Objective:To establish and validate a personalized prediction model of hypopituitarism aftercraniocerebral trauma Methods:A total of 620 patients with craniocerebral trauma who were treated in ourhospital from January 2021 to May 2022 were selected as the research objects They were randomly divided into744modeling group(434 cases)and verification group(186 cases)according to the ratio of 7 3 In the modelinggroup，patients were divided into hypopituitarism group and normal group according to whether they hadhypopituitism;the clinical data of patients were collected，and the risk factors of hypopituitarism aftercraniocerebral trauma were analyzed by single factor and multi factor Logistic regression;an nomogram model wasconstructed to predict hypopituitarism after craniocerebral trauma，the calibration of the nomogram model wasevaluated by calibration curve and H-L goodness of fit test，and the discrimination of the model was verified bydrawing the receiver operating characteristic(OC)curve esults:143 of 434 patients with craniocerebral traumahad hypopituitarism，the incidence was 32 95%;multivariate Logistic regression analysis showed that admission toICU(O=12 644，95%CI 5 800 27 566)，admission GCS score 8 points(O=8 168，95%CI 2 478 26 927)，diffuse cerebral edema(O=5 759，95%CI 2 329 14 241)，cerebral hernia(O=2 220，95%CI1035 4 762)，midline displacement 5 mm(O=13 479，95%CI 6 640 27 360)，increased intracranialpressure(O=6 957，95%CI 2 459 19 682)，skull base fracture(O=2 538，95%CI 1 083 5 950)andlength of stay(O=1 136，95%CI 1 079 1 197)were all independent risk factors for hypopituitarism(P 0.05)The column line graph model was constructed and internally and externally validated，and the results of theH-L goodness-of-fit test showed that 2=7 287，P=0 506 for the modeling group and 2=7 202，P=0 515 forthe validation group，with good agreement;the results of OC curve analysis showed that the area under the OCcurve(AUC)for predicting hypopituitarism in patients with craniocerebral trauma in the modeling and validationgroups were 0 929(95%CI 0 906 0 953)and 0 892(95%CI 0 843 0 942)Conclusion:Admission toICU，GCS score8 points at admission，diffuse brain edema，cerebral hernia，centerline displacement5 mm，increased intracranial pressure，skull base fracture，and length of hospital stay are all risk factors affectinghypopituitarism after craniocerebral trauma，the prediction model based on the above risk factors can effectivelypredict the risk of hypopituitarism after craniocerebral trauma，which is helpful for clinicians to early identifypatients with hypopituitarism after craniocerebral trauma Key wordsbrain trauma;hypopituitarism;influencing factors;nomogram;prediction model颅脑创伤是三大神经系统疾病之一，每年约有5 000万人发生创伤性颅脑损伤1，是全球内致死和致残的主要原因之一，给家庭及社会都带来了沉重的负担2-3。垂体功能减退是颅脑创伤后常见的并发症，患者会出现一系列神经精神症状，但因其临床症状与神经功能障碍容易混杂，无法得到及时准确的治疗，严重影响患者预后4-5。目前颅脑创伤后垂体功能减退的发生机制尚不完全明确，多认为是多种因素作用的结果6。列线图预测模型常用于个体疾病风险预测，在多因素回归分析筛选出有确切预测价值的指标的基础上，再加以整合，可视化展示，有较高的临床指导价值7。因此，本研究对颅脑创伤后垂体功能减退的危险因素进行分析，并构建颅脑创伤后垂体功能减退的风险列线图模型，对降低颅脑创伤后垂体功能减退发生率及改善患者预后具有重要意义。1对象与方法1 1研究对象采用方便抽样法，选取 2021 年 1 月至 2022 年 5月于我院进行治疗的颅脑创伤患者为研究对象。采用Logistic 自变量事件数法计算样本量，确定本研究的样本量为 620 例。将纳入患者按照 7 3 分为建模组(434 例)及验证组(186 例)。将建模组患者根据是否发生垂体功能减退分为减退组(143 例)和正常组(291例)。纳入标准:(1)颅脑创伤患者诊断标准参考颅脑创伤临床救治指南(第三版)8;(2)患者均进行垂体功能检查;(3)患者均自愿参加本研究。排除标准:(1)有脑部外伤史;(2)合并有严重心、肝、肾等脏器严重病变者;(3)合并有其他中枢神经系统疾病者;(4)先天性智力发育障碍;(5)颅脑损伤之前有垂体功能不全者。本研究经我院伦理委员会批准同意。1 2评价指标参考以往研究并