BACKGROUND
The use of indirect methods for estimating reference intervals (RIs) is important in population groups that are difficult to recruit in order to use direct methods to determine them. Several indirect methods are currently available. The percentage of possible pathological data in a database remains unknown.
THE AIM OF THE STUDY
Was to evaluate the possibilities of various mathematical algorithms to calculate the correct boundaries of RI in the presence of a certain proportion of pathological values.
MATERIAL AND METHODS
RIs of the main indicators of CBC for three parameters of the CBC test (leukocytes, hemoglobin, platelets) were studied, using four available methods for indirect assessment of reference intervals: TMC, Bhattacharya, Kosmic, RefineR in 0—3 days newborns.
RESULTS
The possibility of adding pathological data, 10-50% at different distances and in different locations (to the upper, lower border of the RI and on both sides), was assessed relative to the initial boundaries of the RIs in accordance with the of pD and Total Error (+/–1TE) criteria. When using indirect methods for calculating RI, it is necessary to take into account the dependence of the methods on various factors, such as the permissible percentage of pathological data in the database, and the possible difference in the results. Calculations have shown that these properties can vary from parameter to parameter, as well as depending on the characteristics of pathological values in the database.
CONCLUSIONS
In this study, for almost all analytes, it was possible to calculate RIs with at least 10% of pathological data in a database at different distances, which confirms the general trend, and makes it possible to calculate RI with indirect methods.