Bekelis Kimon, Kalakoti Piyush, Nanda Anil, Missios Symeon
Section of Neurosurgery, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, USA.
Department of Neurosurgery, Louisiana State University Health Sciences, Shreveport, Louisiana, USA.
World Neurosurg. 2015 Jul;84(1):82-9. doi: 10.1016/j.wneu.2015.02.032. Epub 2015 Mar 5.
Benchmarking of outcomes and individualized risk prediction are central in patient-oriented shared decision making. We attempted to create a predictive model of complications in patients undergoing benign intracranial tumor resection.
We performed a retrospective cohort study involving patients who underwent craniotomies for benign intracranial tumor resection during the period 2005-2011 and were registered in the National (Nationwide) Inpatient Sample database. A model for outcome prediction based on individual patient characteristics was developed.
There were 19,894 patients who underwent benign tumor resection. The respective inpatient postoperative incidences were 1.3% for death, 22.7% for unfavorable discharge, 4.2% for treated hydrocephalus, 1.1% for cardiac complications, 0.9% for respiratory complications, 0.5% for wound infection, 0.5% for deep venous thrombosis, 2.3% for pulmonary embolus, and 1.5% for acute renal failure. Multivariable analysis identified risk factors independently associated with the above-mentioned outcomes. A model for outcome prediction based on patient and hospital characteristics was developed and subsequently validated in a bootstrap sample. The models demonstrated good discrimination with areas under the curve of 0.85, 0.76, 0.72, 0.74, 0.72, 0.74, 0.76, 0.68, and 0.86 for postoperative risk of death, unfavorable discharge, hydrocephalus, cardiac complications, respiratory complications, wound infection, deep venous thrombosis, pulmonary embolus, and acute renal failure. The models also had good calibration, as assessed by the Hosmer-Lemeshow test.
Our models can provide individualized estimates of the risks of postoperative complications based on preoperative conditions and potentially can be used as an adjunct for decision making in benign intracranial tumor surgery.
结果的基准化和个体化风险预测是面向患者的共同决策的核心。我们试图创建一个良性颅内肿瘤切除术患者并发症的预测模型。
我们进行了一项回顾性队列研究，纳入了2005年至2011年期间接受开颅手术进行良性颅内肿瘤切除术并登记在国家（全国）住院患者样本数据库中的患者。开发了一个基于个体患者特征的结果预测模型。
有19894例患者接受了良性肿瘤切除术。住院患者术后各自的发生率为：死亡率1.3%，不良出院率22.7%，治疗性脑积水发生率4.2%，心脏并发症发生率1.1%，呼吸并发症发生率0.9%，伤口感染发生率0.5%，深静脉血栓形成发生率0.5%，肺栓塞发生率2.3%，急性肾衰竭发生率1.5%。多变量分析确定了与上述结果独立相关的风险因素。开发了一个基于患者和医院特征的结果预测模型，并随后在一个自抽样样本中进行了验证。这些模型显示出良好的区分度，术后死亡风险、不良出院、脑积水、心脏并发症、呼吸并发症、伤口感染、深静脉血栓形成、肺栓塞和急性肾衰竭的曲线下面积分别为0.85、0.76、0.72、0.74、0.72、0.74、0.76、0.68和0.86。通过Hosmer-Lemeshow检验评估，这些模型也具有良好的校准度。
我们的模型可以根据术前情况提供术后并发症风险的个体化估计，并且有可能作为良性颅内肿瘤手术决策的辅助工具。
