急诊非增强头颅CT住院医师与专家报告差异评估
本研究评估英国一家三级医疗中心急诊非增强头颅CT报告中放射科住院医师与神经放射学专家对急性病变判断的差异。共纳入2208例扫描，总体急性病变差异率为4.26%。住院医师对多数急性病变特异性超过99.5%，但对硬膜外血肿、部位不明确颅内出血及梗死的敏感度较低。
要点
- 住院医师对中线移位、脑室内出血、占位效应的敏感度最高，可优先信任其阳性判断。
- 对硬膜外血肿、部位不明确颅内出血、梗死及硬膜下血肿需重点复核，易漏诊。
- 住院医师总体特异性极高，假阳性少见，很少导致不必要治疗。
核心结论
- 放射科住院医师与后续专家报告总体一致性高，支持住院医师参与值班报告。
- 住院医师对关键急性病变诊断特异性高，极少导致不必要治疗。
- 不同病变间敏感度差异明显，可用于指导教学与临床复核以减少漏诊。
研究设计与样本
- 回顾性研究，提取英国一家三级医疗中心放射信息系统内4个月急诊非增强头颅CT全部报告。
- 共纳入2208例扫描，筛选由住院医师出具报告并手工编码急性病变及差异。
- 以神经放射学专家报告为参考标准，评估住院医师报告差异。
关键影像表现/征象
- 急性病变涵盖颅内出血各亚型、占位效应、中线移位、颅骨骨折、急性或亚急性梗死、脑积水及占位性病变。
诊断标准与关键数值
- 任何急性病变的差异共94例，占4.26%，其中38例与颅内出血相关。
- 所有急性病变的特异性均超过99.5%。
- 中线移位敏感度为100%（95%CI 90.3-100.0%），脑室内出血为95.7%（78.1-99.9%），占位效应为92.8%（84.9-97.3%）。
- 硬膜外血肿敏感度为66.7%（9.4-99.2%），部位不明确颅内出血为57.1%（18.4-90.1%），梗死为70%（57.9-80.4%），硬膜下血肿为77.6%（63.4-88.2%）。
临床应用与阅片要点
- 住院医师报告中线移位、脑室内出血及占位效应阳性时可信度高，可减少重复核查。
- 急诊夜班复核应重点查找硬膜外血肿、小量硬膜下血肿及早期梗死征象。
- 对出血部位不明确的病例需与专家讨论或加做进一步影像检查。
局限性
- 单中心研究，结果外推性有限。
- 未提供扫描设备参数及住院医师年资构成。
以上解读依据论文摘要（全文未开放获取或许可不允许翻译转载）。
英文原文（摘要）· 点击展开
AIM: To evaluate the discrepancy rates of acute findings on non-contrast computed tomography head scans (NCHCT) between radiology trainees and neuroradiology consultants at a tertiary care centre in the United Kingdom.
MATERIALS AND METHODS: All reports for NCHCT over a 4 month period conducted in the emergency department were extracted from the radiology information system and screened to identify scans reported by a registrar. Reports were manually coded for acute findings and discrepancies. Acute findings included all subtypes of intracranial haemorrhage (ICH), mass effect, midline shift, cranial fracture, acute/subacute infarct, hydrocephalus, and space occupying lesion.
RESULTS: In total, 2208 scans were included. Discrepancies for any acute finding were found in 94 cases (4.26%), including 38 disagreements related to intracranial haemorrhage. Specificity was over 99.5% for all acute findings. The highest sensitivity was observed for midline shift at 100% (95% confidence intervals 90.3-100.0%), intraventricular haemorrhage at 95.7% (78.1-99.9%) and mass effect at 92.8% (84.9-97.3%). Sensitivity was lowest at 66.7% for extradural haemorrhage (9.4-99.2%) and 57.1 for ICH of unclear location (18.4-90.1%), though the prevalence of these pathologies was low. Sensitivity was also lower for infarcts at 70% (57.9-80.4%) and subdural haemorrhage at 77.6 (63.4-88.2).
CONCLUSION: There is a high concordance between radiology registrar and subsequent consultant reports, supporting registrar on-call reporting. Registrars were highly specific in diagnosing key acute pathologies, meaning few patients would have received unnecessary treatment. Sensitivity varied significantly between pathologies, with certain areas for improvement identified, which can be used to inform local teaching and clinical review to reduce misdiagnoses.
引用
- Nash, Bhebhe, Bow, et al. Where can we improve? An evaluation of discrepancies between radiology resident and consultant reports for noncontrast head computed tomography (NCHCT) scans in the emergency department at a tertiary care centre. Clin Radiol. 2026. PMID: 42392936. doi:10.1016/j.crad.2026.107387
供执业医师工作参考，不替代临床判断；数值与分级以指南原文为准。