Jacobson S R, Speer K P, Moor J T, Janda D H, Saddemi S R, MacDonald P B, Mallon W J
Southern California Center for Sports Medicine, Long Beach, USA.
J Shoulder Elbow Surg. 1995 Nov-Dec;4(6):449-53. doi: 10.1016/s1058-2746(05)80037-0.
The most widely used radiographic classification system for acromial morphology identifies three distinct acromial shapes: type I (flat), type II (curved), and type III (hooked). The purpose of this study was to measure the interobserver and intraobserver reliability of determinations of acromial morphology as defined by this system. Between 1990 and 1992, one hundred twenty-six supraspinatus outlet radiographs were obtained from 126 patients by technicians from Triangle Orthopaedic Associates in Durham, N.C. Six fellowship-trained shoulder surgeons independently reviewed each radiograph and classified it as type I, II, or III on the basis of established guidelines. Two surgeons classified each film a second time in random order. Analysis of variance was performed to obtain coefficients for interobserver and intraobserver reliability. Consensus ratings were then used to classify the 126 radiographs into consensus type I, consensus type II, or consensus type III groups. Percentages of type I, II, and III individual ratings within each consensus group were determined. The intraobserver reliability coefficient was 0.888, interpreted as good to excellent reliability. The interobserver reliability coefficient was 0.516, interpreted as poor to fair reliability. Of the 126 radiographs, 26 (20.6%) were rated as consensus type I, 76 (60.3%) were rated as consensus type II, and 24 (19.1%) were rated as consensus type III. The reliability of observer ratings was lowest when delineation between acromial types II and III was required. The low interobserver reliability makes comparisons of studies by different authors difficult to interpret and obscures the true incidence of acromial morphologic types. It also questions reported correlations between acromial type and shoulder pathologic conditions. It is concluded that a system that incorporates more objective classification criteria and acknowledges the continuous nature of acromial morphologic types may improve interobserver reliability and validate the system's use in making clinical and surgical judgments.
I型（扁平型）、II型（弯曲型）和III型（钩型）。本研究的目的是测量该系统所定义的肩峰形态判定的观察者间和观察者内可靠性。1990年至1992年间，北卡罗来纳州达勒姆市三角骨科协会的技术人员从126名患者那里获取了126张冈上肌出口位X线片。六位接受过专科培训的肩部外科医生独立复查每张X线片，并根据既定指南将其分类为I型、II型或III型。两位外科医生以随机顺序对每张胶片进行了第二次分类。进行方差分析以获得观察者间和观察者内可靠性系数。然后使用一致评级将126张X线片分类为一致I型、一致II型或一致III型组。确定每个一致组内I型、II型和III型个体评级的百分比。观察者内可靠性系数为0.888，解释为可靠性良好至优秀。观察者间可靠性系数为0.516，解释为可靠性差至一般。在126张X线片中，26张（20.6%）被评为一致I型，76张（60.3%）被评为一致II型，24张（19.1%）被评为一致III型。当需要区分II型和III型肩峰时，观察者评级的可靠性最低。观察者间可靠性低使得不同作者的研究比较难以解释，并掩盖了肩峰形态类型的真实发生率。这也对报道的肩峰类型与肩部病理状况之间的相关性提出了质疑。得出的结论是，一个纳入更客观分类标准并承认肩峰形态类型连续性的系统可能会提高观察者间可靠性，并验证该系统在临床和手术判断中的应用。
