肾小球滤过率测量值分布对估计公式中种族系数的影响

Influence of different reference GFR distribution on racial coefficient in GFR estimating equation

目的探讨肾小球滤过率（GFR）估计公式中种族系数是否会受到不同研究中GFR测量值分布不同的影响。方法本研究包括MDRD研究中的1405例白种美籍患者，321例非洲裔美国患者和中国eGFR调查研究中的684例中国CKD患者。首先，白种美籍患者和中国CKD患者的数据库合并，将GFR、年龄（age）、血肌酐（Pcr）进行Iog转换，令Iog[GFR]为应变量，log[Pcr]、log[age]、性别和种族为白变量进行线性回归，计算不加权的种族系数（uRC）；然后按照白种美籍患者的CKD各期比例予中国CKD患者GFR测量值进行加权，使之与白种美籍患者的CKD分期分布一致，计算加权后的种族系数（wRC）。白种美籍患者和非裔美籍患者的数据合库，使非裔美籍患者的CKD分期分布分别与中国患者和白种美籍患者一致，计算相应的种族系数。结果中国患者的uRC为1.197（1.180—1.211），wRC为1.130（1.117—1.143）。与中国患者和白种美籍患者CKD分期一致后，非裔美籍患者的种族系数则分别为1.205（1.191—1.219）和1.233（1.219—1.247）。结论用于开发GFR公式的人群的GFR测量值的分布不同可以导致种族系数的差异。为得到真正的种族系数需要国际合作，应用相同的GFR测量方法和Per测量方法，同时也需要GFR测量值分布一致。

Objective To illustrate if the racial coefficient （RC） be biased by different reference GFR （rGFR） distribution among studies. Methods 1405 white and 321 African- American participants in MDRD study and 684 chronic kidney disease （CKD） patients in Chinese eGFR Investigation Study were included. Firstly, the unweighted datasets of white and Chinese were stacked together, rGFR, age and plasma ereatinine （Per） were log transformed. Linear regression model was constructed using log transformed rGFR as dependent, gender, race and log transformed Pcr and age as independent. Unweighted RC （uRC） for Chinese was calculated. Then, the Chinese CKD distribution of rGFR was weighted to be the same as that in White American, and weighted RC （wRC） for Chinese was calculated. The cases of White- and African-American were stacked together. The cases of African-American were weighted to make the rGFR distribution the same as that in White- American and Chinese population respectively, and RCs for African-American were calculated. Results The nRC for Chinese was 1.197 （1.180-1.211） and the wRC was 1.130 （1.117-1.143）. The two RCs did not overlap with each other. The RCs for African-American were 1.205 （1.191-1.219） and 1.233 （1.219-1.247） respectively. Conclusions The RCs were influenced by the difference of rGFR distribution. To find out the real RC, an international collaborative study is needed, with the same rGFR measure method, strict control of Pcr measurement, and the same rGFR distribution.