不同阅片者对肺亚实性结节临床处理一致性的影响

Discrepancy in film reading of pulmonary subsolid nodules among different readers

目的分析不同放射诊断医生对肺亚实性结节的临床处理方法的一致性及其影响因素。方法回顾49个亚实性肺结节的影像学资料,共有3家医院8位阅片者,包括A单位6位(高级职称A1、A2;中级职称:A3、A4;初级职称:A5、A6)、B与C单位各1位(记为B1、C1)。所有阅片者需要完成的内容包括:结节类型、结节大小、临床处理方法,所有阅片者独立完成阅片,并在1个月后再次进行阅片,并记录上述内容。阅片者间分析采用第一次记录结果。结果临床处理方法的一致性:阅片者A5的内部一致性最低,为0.385(95%CI:0.132~0.638),其次为阅片者A2,为0.502(95%CI:0.226~0.778),其余阅片者的内部一致性均为好;A单位高级、中级、初级职称阅片者间的一致性逐渐降低,前两者各组间的临床处理方法均无统计学差异(均P>0.05),后者有统计学差异(P<0.05);3个单位同年资阅片者间一致性为0.5(95%CI:0.338~0.662),临床处理方法有统计学差异(均P<0.05);通过对临床处理方法出现分歧的结节分析发现,同单位不同年资、不同单位的同年资阅片者间出现分歧的结节分类大部分为部分实性结节患者,不同的是前者对分歧结节的分类一致性较好(Kappa=0.754,95%CI:0.677~0.831),后者对分歧结节的分类一致性较差(Kappa=0.171,95%CI:-0.131~0.473),对于实性大小的测量两者的一致性均较差(Kappa=0.0909,95%CI:-0.088~0.270;Kappa=-0.335,95%CI:-0.677~0.007)。在A单位相同年资的分歧结节均为部分实性结节,对于实性部分的大小,两位阅片者一致性较差(Kappa=-0.849,95%CI:-1.294~1.463)。结论肺结节的临床处理方法、分类以及实性成分的测量在不同阅片者间存在一定的差异,年资、工作单位都可能会影响肺结节的分类、测量以及临床处理策略。

Objective To analyze the discrepancy in film reading of pulmonary subsolid nodules and clinical management among different readers.MethodsThe results of film reading of 49 pulmonary subsolid nodules were retrospectively analyzed.There were 8readers from 3 hospitals,including 6 readers in hospital A（2 senior level readers：A1,A2,2 intermediate level readers：A3,A4,2 junior level readers：A5,A6）,2 readers from hospitals B and C（both were intermediate level：B1,C1）.All readers were asked to independently complete the film reading and to report the nodular type,nodule size and clinical management,and the reading was repeated 1 month late.The intra-reader and inter-reader consistency was analyzed.ResultsThe intra-reader consistency of clinical management varied from 0.385（95%CI：0.132～0.638）to 0.791（95%CI：0.511～1.071）.The intra-reader consistency of clinical management in senior,intermediate and junior readers were decreased with the decrease of seniority in hospital A,and there was no significant difference between senior and intermediate level readers（P〉0.05）,but there was significant difference between intermediate and junior readers（P〈0.05）.The inter-reader consistency of the readers with same seniority from 3 different hospitals was 0.5（95%CI：0.338～0.662）and there was significant difference among them（P〈0.05）.The part-solid nodules were most discrepant in clinical management strategies among the readers in the same hospital with different seniority and in different hospital with the same seniority,and the inter-reader consistency of these nodules in clinical management strategies was good（Kappa=0.754,95%CI：0.677～0.831）for readers in the same hospital with different seniority,but that was poor（Kappa=0.171,95%CI：-0.131～0.473）for readers in the different hospital with the same seniority.The inter-reader consistency of size measurement for part solid was poor（Kappa=0.0909,95%CI：-0.088 to 0.270;Kappa=-0.335,95%CI：-0.677～0.007,respectively）.In hospital A,nodules with discrepant clinical management strategies among same seniority were all part-solid nodules,and the inter-reader consistency of the size measurement for part solid was poor（Kappa=-0.849,95%CI：-1.294～1.463）.Conclusion There are differences in clinical management strategy,classification and measurement of solid components of pulmonary nodules among different readers.The seniority and hospital of readers may affect the results of pulmonary nodule reading.