Non-Invasive Hemoglobin Monitoring Compared to Capillary Blood Sampling in Neonates: A Diagnostic Test Accuracy Review

Abstract

Introduction: Frequent hemoglobin (Hb) monitoring in neonates is often required but painful. Non-invasive spectrophotometry (NIS) devices measure Hb transcutaneously. Their diagnostic accuracy against capillary blood sampling (the current standard) in neonates is unclear.

Methods: Systematic review and bivariate meta-analysis following PRISMA-DTA. Searched PubMed, Embase, IEEE Xplore for studies comparing NIS (Masimo Rad-67, Pronto-7, or equivalent) to capillary Hb in neonates (0–28 days). Quality assessed with QUADAS-2. Primary outcome: bias (mean difference) and limits of agreement (LoA). Secondary: sensitivity to detect anemia (Hb<13 g/dL) and polycythemia (Hb>22 g/dL).

Results: Nineteen studies (n=1,734 paired measurements) were included. Pooled mean bias was –0.32 g/dL (NIS lower than capillary), 95% LoA –1.9 to +1.3 g/dL. Correlation was r=0.78. For anemia detection, sensitivity was 71% (95% CI 62–79%), specificity 88% (81–93%). For polycythemia, sensitivity 68% (54–80%), specificity 91% (85–95%). Bias was larger in infants <30 weeks gestation (–0.65 g/dL) and those with poor perfusion (shock, hypothermia).

Discussion: NIS underestimates Hb relative to capillary samples, particularly in sick or preterm neonates. Wide LoA (±1.6 g/dL) means individual readings are insufficient for transfusion decisions. However, trending over time may be useful. Current devices are not replacements for invasive sampling in unstable infants.

Conclusions: Non-invasive hemoglobin monitoring in neonates has acceptable correlation but insufficient precision for clinical decision-making at Hb thresholds. Use is limited to screening or trend monitoring in stable term infants.