The application of three statistical methods for reference intervals establishment of coagulation assays during pregnancy / 中华检验医学杂志

ABSTRACT

Objective:To compare the practical value of three statistical methods in establishing reference intervals by the indirect method.Methods:This is a methodological evaluation study. The data of coagulation parameters were obtained from laboratory information system, which were from pregnant women who had done prenatal examination in Peking University First Hospital from January 2017 to December 2019. The test results from 32 401 pregnant women were collected. Those healthy pregnant women were divided into three groups early pregnancy group(n=11 151), middle pregnancy group(n=4 872) and late pregnancy group(n=16 378). Statistical analysis was performed for the result of PT, APTT, FIB, TT, D-D and FDP, and the necessity of stratification based on age in different periods of pregnancy was analyzed. Stratification based on age was necessary in three pregnancy groups for PT and APTT, in late pregnancy group for FIB and TT, and in early pregnancy group for D-D. The reference intervals of coagulation parameters were calculated by three statistical methods: non-parametric method, Hoffmann method and Q-Q plot method. Forty-two healthy pregnant women from October 2019 to January 2020 were enrolled as reference individuals for the validation of the reference intervals.The proportions of test results outside the reference intervals in the reference population calculated and compared.Results:The levels of the six coagulation assays vary significantly during the three periods of pregnancy, stratification based on age was necessary in three pregnancy groups for PT and APTT, in late pregnancy group for FIB and TT, and in early pregnancy group for D-D. If the number of test results was large, non-parametric and Hoffmann method provided more similar results, while the reference intervals calculated with Q-Q plot method was slightly wider than Hoffmann method. If the number of test results was small, reference intervals calculated with Hoffmann and Q-Q plot method were more reliable. For pregnant women during early pregnancy under the age of 35, the reference intervals of PT, APTT, FIB and TT calculated by this method were (10.44-13.11)s, (25.29-35.88)s, (2.61-4.64)g/L and (11.53-15.58)s.Conclusion:When establishing the reference interval, stratification according to pregnancy period and age was needed. Hoffmann method can be used as an alternative to the direct method.