基于自主呼吸停止预测的潜在脑死亡供者的早期识别

Early identification of potential brain death organ donors based upon prediction of spontaneous respiratory arrest

目的本研究旨在构建预测神经重症患者在脑损伤后72 h内出现自主呼吸停止的Nomogram模型,用于潜在脑死亡供者的早期识别。方法2017年10月至2019年1月期间,前瞻性纳入中山大学附属第一医院收治的神经重症患者,以创伤性颅脑损伤和自发性脑出血为主,动态记录评估脑损伤的相关参数;其中2017年10月1日至2018年5月31日期间数据用于构建训练集,2018年6月1日至2019年1月31日期间数据用于构建验证集。以脑损伤后72 h内出现自主呼吸停止作为时间兴趣点和分组因素,单因素和多因素分析与自主呼吸停止相关的因素,然后构建Nomogarm预测模型,最后在验证集中对Nomogarm预测模型进行验证。结果依据纳入标准,共127例患者纳入研究;65例进入训练集,62例进入验证集。在训练集中,经单因素和多因素分析显示:中线结构移位(OR=4.56,95%CI:1.87~19.21),环池消失(OR=4.83,95%CI:1.35~16.34),咳嗽反射(OR=3.82,95%CI:1.15~12.42),脑室内出血(OR=3.16,95%CI:1.53~14.52)和血Na^(+)<125 mmol/L(OR=3.06,95%CI:1.53~13.44)是与神经重症患者在脑损伤后72 h内出现自主呼吸停止的相关因素,然后以相关因素构建Nomogram预测模型。在训练集和验证集中,神经重症患者72 h内出现自主呼吸停止率的预测C指数分别为0.81(95%CI:0.76~0.85)和0.80(95%CI:0.75~0.83)。进一步统计分析显示评分140分,160分和170分是危险分界点,其中140分,160分和170分在72 h内自主呼吸停止的概率分别为30.1%,65.6%和93.4%。结论基于脑损伤评估参数构建的Nomogram模型,能预测神经重症患者自主呼吸停止发生的时间,可用于潜在脑死亡供者的早期识别,结果有待进一步外部数据验证。

Objective To construct a nomogram model for spontaneous respiratory arrest prediction in nerocritical patients within 72 hours after brain injury for early identification of potential brain death organ donors.Methods From October 2017 to May 2019,127 hospitalized neurocritical patients(including traumatic brain injury and spontaneous cerebral hemorrhage)were prospectively enrolled and the parameters related to brain injury were dynamically recorded.Among them,the data from October 1,2017 to May 31,2018 were used for constructing the training set and the data from June 1,2018 to January 31,2019 for constructing the validation set.The occurrence of spontaneous respiratory arrest within 72 h after brain injury was regarded as the time interest point and grouping factor.The factors associated with spontaneous respiratory arrest were screened by univariate and multivariate analyses.Then the Nomogarm prediction model was developed and tested in the validation set.Results Sixty-five patients entered the training set and another 62 cases were enrolled into the validation set.In training set,univariate and multivariate analyses indicated that midline shift(OR=4.56,95%CI:1.87~19.21),absent of ambient cistern(OR=4.83,95%CI:1.35~16.34),cough reflex(OR=3.82,95%CI:1.15~12.42),intraventricular hemorrhage(OR=3.16,95%CI:1.53~14.52)and serum Na^(+)<125 mmol/L(OR=3.06,95%CI:1.53~13.44)were associated with spontaneous respiratory within 72 h.In both sets,the predicted C index of spontaneous respiratory arrest rate within 72 h was 0.81(95%CI:0.76~0.85)and 0.80(95%CI 0.75~0.83)respectively.Further statistical analysis implied that 140,160 and 170 points were the dangerous dividing points and these three points were 30.1%,65.6%and 93.4%associated with spontaneous respiratory arrest within 72 h respectively.Conclusions Nomogram model based upon assessment parameters of brain injury may predict the time of spontaneous respiratory arrest in neurocritical patients.It can be used for early identification of potential brain death organ donors.The results require further external data validation.