Premium
Automated screening for tuberculosis by multiparametric analysis of data obtained during routine complete blood count
Author(s) -
Park J.,
Lee H.,
Kim Y.K.,
Kim K. H.,
Lee W.,
Lee K.Y.,
Park Y.J.,
Kahng J.,
Kwon H. J.,
Kim Y.,
Oh E.J.,
Lim J.,
Kim M.,
Han K.
Publication year - 2014
Publication title -
international journal of laboratory hematology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.705
H-Index - 55
eISSN - 1751-553X
pISSN - 1751-5521
DOI - 10.1111/ijlh.12148
Subject(s) - medicine , tuberculosis , population , reproducibility , test set , test (biology) , nuclear medicine , pathology , statistics , mathematics , paleontology , environmental health , biology
Summary Introduction The main goal of this study was to develop a multiparametric cell population data ( CPD ) model that combines information from several morphologic parameters generated by DxH800, in addition to the traditional parameters regularly reported in the CBC ‐diff, and to test the performance of this model in screening the general population for primary tuberculosis ( TB ). Methods A total of 3741 study cases were divided into two groups, test and validation set at cut‐off value of 6000 WBC s/μL. We developed multiparametric model for primary TB screening ( TB hemeprint), selected CPD , and calculated parameters which could discriminate primary TB from other non‐ TB diseases and normal control in test set. We applied it to the validation set, which was a set of completely different samples, to test its reproducibility if applied to a routine laboratory test. Results After screening primary TB using TB hemeprint, sensitivity, specificity, PPV , and NPV were 85.4%, 89.6%, 31.1%, and 99.1%, respectively, in primary TB with lower than 6000 WBC s/μL of test set (test set‐L). In primary TB with higher than 6000 WBC s/μL of test set (test set‐H), those values were 83.1%, 85.6%, 29.7%, and 98.6%, respectively. There were only 0.4% (2/461) and 0.6% (2/326) of normal control samples included in test set‐L and ‐H, respectively. Diagnostic efficiencies except sensitivity in each validation set were very comparable with those in each test set. Conclusion Tuberculosis hemeprint may allow us to screen primary TB with acceptable sensitivity and specificity using combination of TB ‐specific CPD and calculated parameters.