Abstract:

Objective To establish an indirect model based on mathematical statistics method and laboratory information system（LIS） data,and to investigate the reference interval of serum creatinine（Cr） in Ji'an,Jiangxi. Methods By a retrospective analysis of LIS data,9 411 healthy subjects （5 924 males and 3 487 females） from 20 to 90 years old were enrolled. Serum Cr was determined,and the results were analyzed by SPSS 19.0 software to obtain an approximate normal distribution if they were non-normal distribution through datum normality test. The outliers were identified and omitted,while P_2.5-P_97.5 as a reference interval was calculated using non-parametric sorting method and was verified. Results A total of 127 outliers were removed after datum normality test and conversion,and the total number of cases was 9 284. There was statistical significance for serum Cr levels between different age and sex groups（P<0.05）. Therefore,age-specific and sex-specific reference intervals should be established. The reference intervals of serum Cr which had been verified were 65-110 μmol/L for males of 20-59 years old,62-124 μmol/L for males of 60-79 years old,46-79 μmol/L for females of 20-59 years old and 49-85 μmol/L for females of 60-79 years old,respectively. Conclusions The reference intervals of serum Cr based on LIS data are slightly higher than current recommended reference intervals,and they are consistent with the feedback from clinical clients and laboratories,so clinical laboratories should choose,verify or establish suitable reference intervals according to actual situation.

Key words: Creatinine, Biological reference interval, Indirect method