不同方法量化肩关节盂骨缺损比例的信度和效度分析

Analysis of reliability and validity of different methods for quantifying the proportion of shoulder glenoid defect

目的评价基于CT影像的不同量化方法对肩关节盂骨缺损比例评估的信度与效度。方法使用4个肩关节标本,女2例、男2例;死亡年龄分别为55、56、73、50岁。标本采用相同方式制备,均无创伤、骨关节炎和畸形,前期制备过程未出现骨缺损。通过截骨制作关节盂骨缺损模型,每个标本依次截为四个梯度缺损,缺损比例分别约为8%、16%、24%、32%,共计16个骨缺损模型。对16个样本进行实物拍照与CT影像重建,使用表面积法、叠加圆法、Barchilon法、Pico法、Shaha法、Griffith法、Sugaya法、Giles法分别进行骨缺损量化。采用单一测量的双向随机效应模型,计算组内相关系数(interclass correlation coefficient,ICC)评价信度;以表面积法在标本上进行实物测量为参考标准评价效度。结果每种量化方法的一致性ICC均>0.90,具有较高的信度。综合所有骨缺损梯度和影像学图像的结果,表面积法效度最好,为0.83%±0.75%;Barchilon法效度次之,为0.91%±0.93%;叠加圆法与Pico法的效度较好,分别为0.99%±0.87%和1.27%±1.09%Shaha法、Griffith法、Sugaya法的效度较差,分别为6.11%±1.56%、5.06%±1.35%、6.02%±1.61%;Giles法效度最差,为8.40%±3.08%。结论临床实际应用中条件允许的情况下可使用表面积法、叠加圆法量化骨缺损比例;当只能进行线性测量时Barchilon法为首选方法,而Pico法为角度测量的首选方法。

Objective To evaluate the reliability and validity of different quantitative methods based on CT images to evaluate the proportion of shoulder glenoid defect.Methods Four shoulder joint specimens with no trauma,osteoarthritis or deformity were used,including 2 females and 2 males;the average age of death was 58±10 years old;all the specimens were prepared with a standard method with no bone defect occurring before preparation.A glenoid bone defect model was established with each specimen being cut into four defect gradient defects of approximately 8%,16%,24%,and 32% in proportion in sequence.A total of 16 samples were obtained.Physical photography and CT image reconstruction were performed on the 16 samples respectively.A total of 8 quantitative methods were used to quantify bone defects,which were surface area method,superimposed circle method,Barchilon method,Pico method,Shaha method,Griffith method,Sugaya method,and Giles method.Intraclass correlation(ICC)using a consistency model was used to evaluate reliability.Paired t-test was used to evaluate validity,with the physical measurement of the specimens using the surface area method as the reference standard.Result The consistency ICC of each quantitative method was greater than O.9,and all had high reliability.Combining the results of all bone defect gradients and imaging images,the surface area method had the best validity,which was 0.83%±0.75%;the Barchilon method came second,which was 0.91%±0.93%;the superimposed circle method and the Pico method had good validity,which were 0.99%±0.87% and 1.27%±1.09%,respectively;the Shaha method,the Griffith method,and the Sugaya method had poor validity,which were 6.11%±1.56%,5.06%±1.35%,and 6.02%±1.61%,respectively;the Giles method had the worst validity,which was 8.40%±3.08%.Conclusion In clinical practice,surface area method and superimposed circle method are the most reliable to quantify the proportion of bone defect if they can be performed.Otherwise,linear measurement of Barchilon method is the favored method while PICO method is the favored method for angle measurement.