OBJECTIVE: To analyze the impact of specific comorbidities on the operation-specific and overall surgical success of LTR in a large multicenter cohort and validate a predictive model for surgical success.
METHODS: A large retrospective multicenter 10-year review was undertaken to validate the data of a previous single-center study (Wertz et al. Laryngoscope 2020) which identified specific predictive comorbidities which impacted LTR outcomes. A Monte Carlo simulation based on the previous data set suggested that 300-400 cases would be needed to optimize the statistical power of a Bayesian model developed from the single-center data to predict surgical success. An IRB-approved data-sharing agreement was executed for 4 large U.S.
METHODS: A virtual REDCap® data entry form inquired about patient characteristics that best predicted surgical success in the single-center model. These included demographics, surgical approaches, cardiac, airway, genetic, endocrine, musculoskeletal, gastrointestinal, and pulmonary comorbidities; details of the surgical procedures, and results of esophagogastroduodenoscopy (EGD), esophageal pH/impedance and flexible bronchoscopy with bronchioalveolar lavage (BAL) were included. Surgical success defined as successful decannulation or resolution of airway symptoms was recorded as single surgery success and overall success following open surgical revision surgery. Multivariate Bayesian analysis, logistical regression, and Kaplan-Meier analysis were performed.
RESULTS: 542 patients were identified, including 165 from the single-center study and an additional 377 patients from the multicenter group. The median age was 36 months at the time of the most recent surgery. 70.9% of the LTRs were double-staged procedures. The overall success rate was 86.4% and operation-specific success rate was 69.2%. The specific comorbidities and aerodigestive test results that impacted success based on univariate analysis included staging, bronchiectasis, pulmonary hypertension, GERD, ASD, PDA, grade of stenosis, advanced levels of stenosis, Trisomy 21, MRSA, prior open surgery at another hospital, and gross appearance on EGD. Bayesian model averaging with backward selection was used to validate and refine a predictive model for surgical success with favorable receiver operating curve characteristics - AUC values of 0.827 for single surgery success and 0.797 for overall success.
CONCLUSIONS: With over 500 patients reviewed, this was the largest multicenter study of LTR to date, which elucidated the impact of comorbidities on success with LTR and was able to improve upon the predictive modeling based on single-center data. Patient factors are most critical in the outcome of LTR. Stage and levels of stenosis, as well as pulmonary and GI conditions most strongly impact the likelihood of success. Future prospective case-control studies will be performed to further optimize the current model for outcome prediction and patient management.
目的:在大型多中心队列中分析特定合并症对LTR手术特异性和整体手术成功率的影响,并验证手术成功率的预测模型。
方法:进行了一项大型回顾性多中心10年审查,以验证先前单中心研究的数据(Wertz等人。喉镜2020)确定了影响LTR结果的特定预测性合并症。基于先前数据集的蒙特卡洛模拟表明,需要300-400例病例来优化从单中心数据开发的贝叶斯模型的统计能力,以预测手术成功。为4个大型美国
方法执行了IRB批准的数据共享协议:虚拟REDCap®数据输入表询问了患者特征,这些特征最好地预测了单中心模型中的手术成功。这些包括人口统计,手术方法,心脏,气道,遗传,内分泌,肌肉骨骼,胃肠,和肺合并症;外科手术的细节,和食管胃十二指肠镜检查(EGD)的结果,包括食管pH/阻抗和支气管肺泡灌洗(BAL)的柔性支气管镜检查。定义为成功拔管或解决气道症状的手术成功被记录为单次手术成功和开放式手术翻修手术后的总体成功。多元贝叶斯分析,后勤回归,进行Kaplan-Meier分析.
结果:确认了542例患者,包括单中心研究的165例患者和多中心组的377例患者.在最近一次手术时,中位年龄为36个月。70.9%的LTR是双阶段程序。总体成功率为86.4%,手术成功率为69.2%。根据单变量分析,影响成功的具体合并症和呼吸消化测试结果包括分期,支气管扩张,肺动脉高压,GERD,ASD,PDA,狭窄程度,狭窄的晚期水平,21三体MRSA,之前在另一家医院做了开放性手术,和在EGD上的粗略外观。使用带有反向选择的贝叶斯模型平均来验证和完善手术成功的预测模型,具有良好的受试者工作曲线特征-单次手术成功的AUC值为0.827,总体成功的AUC值为0.797。
结论:回顾了500多名患者,这是迄今为止最大的多中心LTR研究,阐明了合并症对LTR成功的影响,并能够改进基于单中心数据的预测建模。患者因素在LTR的结果中最为关键。狭窄的分期和水平,以及肺部和胃肠道疾病对成功的可能性影响最大。将进行未来的前瞻性病例对照研究,以进一步优化当前的结果预测和患者管理模型。