Objective: Near infrared spectroscopy (NIRS) failed till now to support clinical decisions in neonates. One main problem is the low reproducibility of measurements. The purpose of the present study was to introduce quality criteria to increase this low reproducibility.

Methods: In a prospective cohort study in 40 neonates repeated measurements on the calf with NIRS and venous occlusion were performed. In each neonate the NIRS optodes were reapplied five times. At each reapplication five measurements at rest were performed. Tissue oxygenation index (TOI), mixed venous oxygenation (SvO₂), fractional oxygen extraction (FOE), haemoglobin flow (Hbflow), oxygen delivery (DO₂), and oxygen consumption (VO₂) were assessed. Heart rate and arterial oxygen saturation (SaO₂) were measured by pulse oximetry. NIRS measurements with linear changes during venous occlusions with r² >0.95 were included for further analysis (first quality criterion). The second quality criterion was: 0 ⩽ TOI (SaO₂–(FOE*100)) ⩽ (FOE*100)*0.2. Analysis of variance components were applied to the data after the introduction of the quality criteria.

Results: With the introduction of the quality criteria the variance components of reapplications and measurements decreased in all parameters: TOI from 46.6% to 32.0%, SvO₂ from 76.8% to 33.0%, FOE from 73.1% to 31.2%, Hbflow from 70.3% to 48.3%, DO₂ from 71.5% to 49.2% and VO₂ from 70.9% to 63.6%.

Conclusion: With the application of the quality criteria peripheral NIRS measurements have reached a decrease in retest variability and an increase of reproducibility that these measurements may support clinical decisions in future. We suggest using these quality criteria in other study populations and with other devices.