儿童头部外伤后CT检查决策规则的适用性

摘要

Objective To explore the applicability of the three commonly used CT examination decision rules in Chinese head injured children. Methods This prospective observational study included 1538 children and adolescents (aged < 18 years), who were treated at the Emergency Department of First Hospital of Shanxi Medical University after head injuries. The three clinical decision rules include the Children's Head Injury Algorithm for the Prediction of Important Clinical Events (CHALICE; UK); the prediction rule for the identification of children at very low risk of clinically important traumatic brain injury, that was developed by the Pediatric Emergency Care Applied Research Network (PECARN; USA), and the Canadian Assessment of Tomography for Childhood Head Injury (CATCH) rule. Diagnostic accuracy had been evaluated by using the rule-specific predictor variables to predict each rule-specific outcome measure in populations who met inclusion and exclusion criteria for each rule. Sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and ROC curve were referred to the diagnostic accuracy. Indicators were characterized by 95％ CI. Results Of the 1538 patients, CTs were obtained for 339 patients (22.04％). Forty-nine patients (3.19％) had positive CT results, 8 patients (0.52％) underwent neurosurgery, 2 patients (0.13％) died, and 1 patient (0.07％) may be missed. In this study, CHALICE was applied for 1394 children (90.70％; 95％ CI: 89.24％-92.15％), PECARN for 801 children (52.11％; 95％ CI: 49.62％-54.61％), and CATCH for 325 patients (21.15％; 95％CI: 19.10％-23.19％). The validation sensitivities of CHALICE, PECARN, and CATCH rules were 92.6％(74.2％-98.7％), 100％ (56.1％-100％), and 85.7％ (42.0％-99.2％), respectively; the specificities were 78.1％(75.7％-80.2％), 48.0％ (44.5％-51.5％) and 70.8％ (65.4％-75.6％); positive predictive value were 7.7％ (5.1％-11.3％), 0.9％ (0.4％-1.9％) and 6.1％ (2.5％-13.2％); and negative predictive value were 99.8％ (99.2％-100％), 99.1％ (98.1％-99.6％), and 99.6％ (97.2％-100％), respectively. Conclusions The clinical decision rules of CHALICE, PECARN and CATCH have high sensitivities. The specificity of PECARN rule is lower than those of CHALICE and CATCH rules. The above three clinical decision rules can be used for the decision of CT examination in Chinese children with head injury in practice.%目的 探讨国外常用的三种儿童头部外伤后CT检查决策规则在我国的适用性.方法 本前瞻性观察研究纳入了1538例因头部外伤就诊于山西医科大学第一医院急诊医学中心的未成年人(年龄<18岁).三种儿童头部外伤后CT检查决策规则包括:英国的儿童头部损伤重要临床事件预测算法(children's head injury algorithm for the prediction of important clinical events,CHALICE)、美国儿科急诊护理应用研究网研究的低风险颅脑损伤的头部外伤未成年人头颅CT检查的临床决策规则(paediatric emergency care applied research network,PECARN)以及加拿大儿童头部损伤断层扫描评估(Canadian assessment of tomography for childhood head injury,CATCH).本研究在满足各规则的纳入、排除标准的患者人群中,以各决策规则的观察结局作为金指标,评估上述三种规则预测变量的诊断准确性.诊断准确性指标评估包括:敏感性,特异性,阴性预测值(NPV),阳性预测值(PPV)和ROC曲线,以95％置信区间表示.结果 在1538例患儿中,共339例(22.04％)患儿接受了头颅CT扫描.其中,49例(3.19％)CT扫描阳性,8例(0.52％)患者接受了神经外科手术,2例患儿(0.13％)死亡,1例(0.07％)患儿可能被漏诊.本研究中,CHALICE适用于1394例患儿(90.70％,95％CI 89.24％～92.15％),PECARN适用于801例(52.11％,95％CI 49.62％～54.61％),CATCH则适用于325例患儿(21.15％,95％CI 19.10％～23.19％).CHALICE、PECARN、CATCH规则的敏感性分别为:92.6％(74.2～98.7)、100％(56.1～100)及85.7％(42.0～99.2);特异度分别是:78.1％(75.7～80.2)、48.0％(44.5～51.5)及70.8％(65.4～75.6);阳性预测值为:7.7％(5.1～11.3)、0.9％(0.4～1.9)及6.1％(2.5～13.2);阴性预测值为:99.8％(99.2～100)、99.1％(98.1～99.6)及99.6％(97.2～100).结论 CHALICE、PECARN、CATCH临床决策规则均具有高度敏感性,较CHALICE、CATCH规则而言,PECARN规则的特异度较低.