Diagnoctic test accuracy of nucleated red blood cells in non-hematologic intensive care patients (meta-analysis)

Background. Nucleated red blood cells (NRBC) are reticulocyte precursors detected in the complete blood cell count and absent in the bloodstream of a healthy adult. The presence of NRBC in adult non-hematologic patients is being studied as a biomarker of an unfavorable outcome. Objective. To determine the NRBC-positive patients’ prevalence in the non-hematologic intensive care unit (ICU) and assess NRBC diagnostic accuracy as a predictor of mortality. Material and methods. A meta-analysis that included 19 studies with 5785 severe and critically ill patients was performed: 16 articles contained qualitative data of NRBC and 5 studies had data of NRBC quantitative assays. The literature search was conducted in the following databases — PubMed, Researcher Gate and e-library. Risk of bias in non-randomized studies was assessed using the Russian-language version of the Newcastle Ottawa Scale, studies included in meta-analysis of diagnostic test accuracy additionally was evaluated using the Russian-language version of the QUADAS Questionnaire. Results. The pooled prevalence of NRBC-positive non-hematologic ICU patients was 36.6%. The overall mortality of NRBC-positive patients was 49.0%. The weighted mean difference of NRBC values between survivors and non-survivors was –520.43 μl–1, for patients with ARDS: –1093.11 μl–1, and without ARDS: –282.87 μl–1. A meta-analysis of studies with univariate analysis showed that the pooled risk ratio (RR) of intrahospital all-cause mortality increased 4.28-fold (95% CI 2.63—6.94, p<0.00001, I2=96%, p<0.00001) in NRBC-positive patients, in studies with multivariate analysis (Cox Regression) the pooled hazard ratio (HR) of all-cause mortality was 1.26 (CI 95% 1.00—1.59, p=0.05, I2=65%, p=0.04). Hierarchical summary receiver operating characteristic analysis showed areas under the curves 0.745, pooled sensitivity and specificity were 72.6% and 79.5% respectively, and diagnostic test accuracy was 73.7%. Conclusion. The presence of nucleated red blood cells in the complete blood count test with absent of hematopathology associated with severe hypoxia and/or increased release of cytokines, and can be used as a mortality predictor.