Stoddart Michael, Pearce Oliver, Smith James, McCann Philip, Sheridan Barnaby, Al-Hourani Khalid
Trauma and Orthopaedics, Royal National Orthopaedic Hospital, London, GBR.
Trauma and Orthopaedics, Southmead Hospital, Bristol, GBR.
Cureus. 2020 Jun 9;12(6):e8520. doi: 10.7759/cureus.8520.
Several classifications for proximal humeral fractures exist, with excellent reliability and reproducibility of such classifications being a desirable feature. Despite their widespread use, these systems are variable in both reliability and accuracy. We aimed to, a) assess and compare the reliability of the Neer (complete and abbreviated versions) and Arbeitsgemeinschaft für Osteosynthesefragenbeing (AO) classifications, and b) identify if computed tomography (CT) made any difference to the reliability of Neer and AO classifications when compared to plain radiographs alone. Materials and methods: This is a single-centre retrospective study identifying all proximal humeral fractures presenting between February 2016 and February 2017 as a result of trauma that subsequently proceeded to CT. Two specialty orthopaedic trainees analysed the plain radiographs as well as CT images over two rounds, spaced two weeks apart. The Neer 16-grade, abbreviated Neer 6-grade and AO classifications were used. Intra- and inter-observer reliability of each classification system was assessed using the Kappa coefficient. Results: Twenty-two patients were included. The mean age was 62 years (SD 14.5). Management changed in 9/22 patients based on CT. Computed tomography changed Neer-16 type in 16% observations, Neer-6 in 10%, and AO in 23%. This was significant when comparing Neer-6 and AO classifications (p = 0.04). Neer-6 had the best inter-observer reliability (0.737) with the management of one patient changing after CT. On X-ray and CT, intra-observer agreement was substantial, >0.7, using Neer-16 and Neer-6 (p<0.005). Inter-observer agreement for Neer-16 and Neer-6 was substantial, >0.7 (p<0.005). In comparison, intra- and inter-observer agreements for AO were lower on X-ray and CT, 0.4-0.6, (p<0.005).
Our study shows that simplicity is key with a high degree of reliability in the abbreviated Neer classification. Computed tomography allowed greater reliability than radiographs in classifying fractures, affecting management decisions in 41% of patients. The comprehensive Neer classification showed similar intra- and inter-observer reliabilities to AO.
肱骨近端骨折存在多种分类方法，理想的分类方法应具有高度的可靠性和可重复性。尽管这些分类方法被广泛应用，但它们在可靠性和准确性方面存在差异。我们旨在：a）评估和比较Neer（完整版和简化版）及AO（Arbeitsgemeinschaft für Osteosynthesefragenbeing）分类的可靠性；b）确定与单纯X线片相比，计算机断层扫描（CT）对Neer和AO分类的可靠性是否有影响。材料与方法：这是一项单中心回顾性研究，纳入2016年2月至2017年2月因创伤导致肱骨近端骨折且随后接受CT检查的所有患者。两名骨科专科实习医生分两轮分析X线片和CT图像，两轮间隔两周。采用Neer 16级、简化Neer 6级和AO分类。使用Kappa系数评估每个分类系统的观察者内和观察者间可靠性。结果：共纳入22例患者。平均年龄62岁（标准差14.5）。9/22例患者基于CT改变了治疗方案。CT改变Neer-16型分类的占16%，改变Neer-6型分类的占10%，改变AO分类的占23%。比较Neer-6和AO分类时差异有统计学意义（p = 0.04）。Neer-6具有最佳的观察者间可靠性（0.737），1例患者在CT检查后改变了治疗方案。在X线片和CT上，使用Neer-16和Neer-6时观察者内一致性较高，>0.7（p<0.005）。Neer-16和Neer-6的观察者间一致性较高，>0.7（p<0.005）。相比之下，AO在X线片和CT上的观察者内和观察者间一致性较低，为0.4 - 0.6（p<0.005）。结论：我们的研究表明，简化版Neer分类的关键在于简单且具有高度可靠性。在骨折分类方面，CT比X线片具有更高的可靠性，41%的患者治疗决策受其影响。完整版Neer分类的观察者内和观察者间可靠性与AO相似。
