骨龄测定的一致性评价

Concordant Evaluation of Skeletal Age Assessment

目的评价骨龄(skeletal age,SA)测定中单个骨骺核观察者间人工评级一致性及总体一致性。方法本研究选择2006年3月至2009年5月在四川大学华西第二医院确诊的中枢性性早熟患儿共计128例为研究对象。将其随机分为病例组(n=61)和对照组(n=67)。其中6例女性患儿为8岁时乳房发育者。2位医师盲法用TW2法桡骨、尺骨和短骨(radius ulna and short bones,RUS)骨龄中国标准,回顾性分析患儿治疗前左手腕部X射线片,进行人工评级并计算骨龄。采用SPSS 13.0统计学软件计算Kappa值,并进行u检验。结果①13个骨骺核的Kappa值分别为:桡骨(0.812±0.051)、尺骨(0.944±0.022)、第一掌骨(0.925±0.028)、第三掌骨(0.703±0.050)、第五掌骨(0.789±0.042)、第一近节指骨(0.964±0.021)、第三近节指骨(0.941±0.026)、第五近节指骨(0.887±0.034)、第三中节指骨(0.917±0.031)、第五中节指骨(0.919±0.030)、第一远节指骨(0.772±0.050)、第三远节指骨(0.883±0.040)、第五远节指骨(0.856±0.041)。对13个Kappa值进行u检验,P值均小于0.05。②总体一致性:Kappa=0.776,标准误为0.044(u=16.128,P<0.05)。结论骨龄测定中单个骨骺核观察者间人工评级的一致性较高。实际工作中,应通过定期培训,提高评价观察者之间的一致性,修正诊断结果,提高诊断准确性。

Objective To evaluate the concordance of artificial rank with single epiphyseal nucleus in skeletal age（SA） assessment between observers and the total concordance.Methods From March 2006 to May 2009,128 children with central precocious puberty were admitted to the West China Second University Hospital.They were divided into case group（n=61） and control group（n=67）.In control group,6 girls started breast development at age 8.Left hand-wrist X-ray radiographs before therapy were retrospectively blindly analyzed by two doctors with radius,ulna and short bones（RUS） skeletal age of TW2 method. Then the two doctors ranked artificially and calculated skeletal age.Kappa values were calculated and u test was performed in SPSS 13.0.Results①The Kappa values of the 13 epiphyseal nucleuses were：Radial bone（0.812±0.051）,ulnar bone（0.944±0.022）,first metacarpal bone（0.925±0.028）,metacarpal bone （0.703±0.050）,fifth metacarpal bone（0.789±0.042）,first proximal phalanx（0.964±0.021）,proximal phalanx（0.941±0.026）,fifth proximal phalanx（0.887±0.034）,middle phalanx（0.917±0.031）,fifth middle phalanx（0.919±0.030）,first distal phalanx（0.772±0.050）,distal phalanx（0.883±0.040）, fifth distal phalanx（0.856±0.041）.For the 13 Kappa values,u test was performed and P values were all smaller than 0.05.②Total concordance：Kappa = 0.776 and standard error = 0.044（u = 16.128,P0.05）.Conclusion The concordance of artificial rank with single epiphyseal nucleus in skeletal age assessment between observers is high.In practice work,concordance between observers is developed by regularly training,then diagnostic results are correlated and accuracy is advanced.