Abstract:

Objective Using indirect method,to establish the reference intervals of total bilirubin（TB） and direct bilirubin（DBil）,to compare with the reference interval of industry standard（direct method）,and to verify its reliability. Methods The data of healthy subjects were collected from January 2017 to December 2017. There were 52 589 males and 36 819 females for TB data,and there were 51 555 males and 36 228 females for DBil data. Using skewness-kurtosis test,the normality of data was analyzed. If not the original data were transformed through BOX-COX transformation to obtain an approximate normal distribution,outliers were identified and omitted by turkey method. The indirect reference intervals were taken by Hoffmann method and compared with the reference interval of direct method. Results The results of skewness-kurtosis test showed that TB and DBil data were non-normal distribution. Using BOX-COX transformation to transform data into normality,784 males and 578 females of TB data and 925 males and 675 females of DBil data were eliminated by turkey method after transformation. There was statistical significance for TB and DBil between males and females（P=0.000）. Therefore,it is necessary to establish reference intervals according to different sex. Using Hoffmann method,the reference intervals were TB male≤22.4 μmol/L,TB female≤18.5 μmol/L,DBil male≤6.4 μmol/L and DBil female≤5.1 μmol/L. The relative deviation between TB male,TB female and DBil male reference intervals and direct method reference interval was lower than reference change value（RCV）. The relative deviation between DBil female reference interval and direct method reference interval was higher than RCV. Conclusions The feasibility of using indirect method to establish The reference intervals of TB and DBil is verified.

Key words: Total bilirubin, Direct bilirubin, Indirect method, Reference interval