Neer分型的可靠性研究

Interobserver reliability and intraobserver reproducibility of Neer classification system

目的通过比较榆查者Neer分型的组间一致性与组内可藿复性评价其可信度和可重复性，以及经验对Neer分型的影响。方法选取2010年1月至2010年12月收治的40例肱骨近端骨折患者的影像学资料（术前肩关节正位、侧位、改良腋位X线片，以及肩关节横断位CT平扫或三维重建），分2次（至少间隔3个月）呈递给12名检查者，以同样的流程进行16分法Neer分型，使用Kappa检验评估组间可信度和组内可重复性。同时比较接受过肩关节外科々科培训的检查者（专业组，6人）与未接受过肩关节外科专科培训的检查者（对照组，6人）的分型一致性，并重组数据后评估简化分型的可靠性。结果2次测试Neer分型组间可信度分别为0．534和0．473，组内可重复性为0．669。专业组分型可靠性显著高于对照组，差异有统计学意义（P〈0．05）。所有数据在新的6分法（简化分型）基础上重组后的组间可信度为0．581和0．502，组内可重复性为0．680，组问可信度有提升，但一致性仍维持在一般水平。所有12名检查者分型一致的病例占总病例数的17．5％（第1次测试）和15．O％（第2次测试）。结论本研究对Neer分型评估取得了一般的组间可信度和良好的组内可重复性，并显示专业水平是分型可靠性的重要影响因素，肩关节外科专业培训对提高组间可信度和组内可重复性至关重要；简化Neer分型对提升可靠性并无明显帮助。

Objective To assess the interobserver reliability and intraobserver reproducibility of Neer classification system and its influencing factors. Methods The present study reviewed the series preoperative radiographs （including those of scapular anteroposterior view, scapular lateral view and modified Velpeau axillary view and an axial CT scan） of 40 patients who had been treated in our department from January＇ 2010 to December 2010 for pruximal humeral fractures. The radiographs were assessed by 12 individual observers on 2 separate occasions with an interval of 3 months at least. Half of the observers （the professional group; n = 6） had received a shoulder fellowship training and the other half （the control group; n = 6） had not. All the observers were asked to categorize the radiographs according to the Neer classification system of 16 types of fractures in a same process. The reliability and reprodueibility of the system were assessed with the Kappa statistics. Com- parisons of classification agreement were made between the professional group and the control group. We also evaluated the simplified Neer system of only 6 types of fi＇aetures with recombinant data. Results The in- terobserver reliability coefficients were 0. 534 and O. 473 for the first and second assessments, with an intraob- server reproducibility coefficient of O. 669. The agreement level in the professional group was significantly higher than in the control group （ P 〈 0.05）. The interobserver reliability coefficients of the simplified Neer system were 0.581 and O. 502, with an intraobserver reproducibility coefficient of 0. 680. Use of the modified Neer system did not elevate the agreement level beyond the moderate range. The classification was agreed on by all the observers in 17.5% of the fractures during the first assessment and in 15.0% during the second assessment. Conclusions Neer classification nmy have fair interobserver reliability and moderate intraobserver reproducibility. Experience of shoulder fellowship training is an important factor inlluencing the reliability of the Neer system. Simpli- fication of the system may not help increase its reliability.