Evaluation of point-of-care complete blood count and C-reactive protein testing for routine use in a pediatric emergency department.

BACKGROUND

Point-of-care testing (POCT) of complete blood count (CBC) and C-reactive protein (CRP) may support rapid decision-making in pediatric emergency care. We evaluated the performance of the Horiba Microsemi analyzer operated by clinical staff in the emergency department (ED) of V. Buzzi Children's Hospital (Milan, Italy) compared with laboratory testing.

METHODS

An unselected cohort of 195 patients (108 males and 87 females; median age 6 years, IQR 2-11 years) admitted to the ED was enrolled over 6 weeks to reflect routine clinical practice. Capillary and venous samples were analyzed using the Horiba Microsemi CRP LC-767G and compared with simultaneously collected venous samples analyzed by laboratory reference systems (Sysmex XN-9000 for CBC and Abbott Alinity for CRP). Method comparison was performed using MedCalc Statistical Software, applying Passing-Bablok regression and Bland-Altman analysis. Accuracy was assessed by comparing mean bias with European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) biological variation-based performance specifications.

RESULTS

Strong correlation (Spearman correlation coefficient, ρ ≥ 0.90) between POCT (capillary and venous) and laboratory measurements was observed for mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), and C-reactive protein (CRP). For lymphocyte (LYMPH) and granulocyte (GRAN) counts, a strong correlation was observed only in venous samples. Moderate correlation (0.75 ≤ ρ < 0.90) was observed for white blood cells (WBC) and platelets (PLT) in both matrices, for hemoglobin (HGB) in venous samples only, and for lymphocytes (LYMPH) and granulocytes (GRAN) in capillary samples. Poor correlation (ρ < 0.75) was found for red blood cells (RBC), hematocrit (HCT), mean corpuscular hemoglobin concentration (MCHC), and monocytes (MONO) in both matrices, and for HGB in capillary samples. According to EFLM specifications, MCHC, LYMPH, and CRP met the predefined desirable performance specifications in both matrices, whereas WBC and GRAN met these criteria only in venous POCT samples. Significant bias was observed for RBC parameters and PLT.

CONCLUSION

Cautious interpretation is warranted for WBC, and particularly for RBC parameters and PLT. The observed data dispersion may reflect pre-analytical variability, potentially affecting the precision of the findings and the applicability of POCT in routine emergency practice, especially for capillary samples.