用累积和控制图建立早孕期胎儿颈项透明层厚度测量的质控体系

目的采用累积和控制图（CUSUM）对早孕期胎儿颈项透明层厚度（nuchal translucency，NT）的测量值进行质控分析，建立NT测量的质控体系。 方法选取2018年1月至2019年1月广州市妇女儿童医疗中心以中国南方人群的胎儿NT中位数值作为参考，将本中心的NT测量值转换为NT中位数倍数值（MoM值），计算每位操作者每天的NT MoM值的中位数和SD值，用CUSUM控制图建立NT测量质控系统。对一位经验丰富的超声医师近20 d的NT测量数据进行评估，将失控的病例与在控病例进行比较分析，检验本质控体系的有效性。从第一位超声医师的NT测量数据中，选取累积和（cumulative sum）为0的该日的所有NT数据，人为的将该日所有NT值±1.0 mm及±0.5 mm，进行CUSUM质控分析，检验质控体系的有效性。采用本质控系统对1位初学者的NT测量结果进行分析，并将所有入组NT图像进行评分，验证CUSUM控制图的有效性。不同组NT图像评分结果之间的比较采用χ²检验。 结果本中心早孕期胎儿NT MoM的中位值为1.1 MoM，SD为0.23 MoM，建立CUSUM控制图。对1位经验丰富的超声医师的近20 d的NT测量结果进行质控分析，结果提示第8、13天的NT测量值出现明显偏倚，通过对这2 d所有的NT图像进行评分，发现图像满意率仅38.9%和40%。当NT测量数值误差超过±1.0 mm时，本CUSUM控制图可以发现数据偏倚。对第2位超声医师的数据进行质控分析，发现数值明显偏小，此结果与图像评分结果一致。 结论CUSUM控制图对NT测量质量进行评估是有效的超声质控方法。

Cumulative sum plots in first-trimester nuchal translucency quality assurance

ObjectiveTo evaluate the quality assurance of fetal nuchal translucency (NT) in the first trimester by cumulative sum (CUSUM) plots. MethodsBased on the reference range of median NT, the median of the NT MoM distribution and standard deviation (SD) were determined. CUSUM plots were created to assess whether there was an over- or under-estimation of NT measurement for each sonographer. An NT image scoring system was established to testify the results analyzed by CUSUM plots. ResultsThe median NT MoM at our center was 1.1 mm, and SD was 0.23mm. CUSUM plots were created for NT assessment. For an experienced sonographer, 20 days of NT measurement were included for CUSUM scoring, which showed that NT on days 8 and 13 was overestimated. We then investigated all the NT images on these two days by the NT image scoring system, and the result was in agreement with CUSUM plots. When the measurement error of NT was above 1.0 mm, the median NT MoM lied outside the accepted limits by CUSUM plots. For a beginner sonographer, the results obtained by CUSUM plotswere quite similar to those obtained by the image scoring system. ConclusionThe application of cumulative sum plots in quality assurance of nuchal translucency measurement is acceptable.