Department of Neurosurgery, University of Cincinnati, Cincinnati, Ohio, USA.
Neurosurgery. 2012 Dec;71(6):1064-70; discussion 1070. doi: 10.1227/NEU.0b013e31826f7c16.
The epidemiology of traumatic brain injury (TBI) is often studied through the use of International classification of disease, ninth revision, clinical modification (ICD-9-CM), diagnosis codes from the Centers for Disease Control and Prevention TBI Surveillance System. Recent studies suggest that these codes may underestimate the burden of TBI because of inaccuracies and low sensitivity.
To determine the sensitivity and specificity of ICD-9-CM codes in a severe TBI population.
We retrospectively reviewed medical records of all hospital admissions including computed tomography of the head at a single center to identify severe blunt TBI patients, their injuries, and the neurosurgical procedures performed. We calculated sensitivity and specificity by comparing ICD-9-CM diagnosis and procedure codes assigned by hospital coders with medical records, the gold standard.
In 2008, there were 148 qualifying admissions. These codes were 89% sensitive for the presence of any severe TBI. However, one-fifth of these cases were identified only with a code defining a nonspecific head injury. Next, we studied types of TBI by categories defined by the Centers for Disease Control and Prevention (morbidity groups) and by ICD-9-CM codes for types of injury (any skull fracture, intracranial contusion, intracranial hemorrhage, concussion/loss of consciousness) and found widely varying sensitivity and specificity for both. In general, these codes had higher specificity than sensitivity. Both sensitivity and specificity were > 80% for only 2 categories: any skull fracture and intracranial hemorrhage. In contrast, we found high sensitivity and specificity for neurosurgical procedures (97% and 94%).
ICD-9-CM codes were sensitive for the presence of any severe TBI, but further classification of specific types of TBI was limited by variable sensitivity/specificity. Use of these codes should be supplemented by other methodology.
创伤性脑损伤 (TBI) 的流行病学研究通常通过使用国际疾病分类，第九修订版，临床修正 (ICD-9-CM)，疾病控制和预防中心 TBI 监测系统的诊断代码来进行。最近的研究表明，由于不准确和低灵敏度，这些代码可能低估了 TBI 的负担。
确定 ICD-9-CM 代码在严重 TBI 人群中的灵敏度和特异性。
我们回顾性地审查了单一中心所有住院患者的病历，包括头部计算机断层扫描，以确定严重钝性 TBI 患者、他们的损伤以及进行的神经外科手术。我们通过将医院编码员分配的 ICD-9-CM 诊断和程序代码与病历（金标准）进行比较来计算灵敏度和特异性。
2008 年，有 148 例符合条件的入院患者。这些代码对于任何严重 TBI 的存在具有 89%的敏感性。然而，其中五分之一的病例仅通过定义非特异性头部损伤的代码来识别。接下来，我们通过疾病控制和预防中心 (发病率组) 定义的类别以及用于损伤类型的 ICD-9-CM 代码 (任何颅骨骨折、颅内挫伤、颅内出血、脑震荡/意识丧失) 研究了 TBI 的类型，发现这两种方法的灵敏度和特异性都有很大差异。一般来说，这些代码的特异性高于敏感性。只有 2 个类别具有 >80%的灵敏度和特异性：任何颅骨骨折和颅内出血。相比之下，我们发现神经外科手术的灵敏度和特异性都很高 (97%和 94%)。
ICD-9-CM 代码对于任何严重 TBI 的存在都是敏感的，但特定类型 TBI 的进一步分类受到灵敏度/特异性变化的限制。应使用其他方法补充这些代码。
