Reproducibility of expired breath hydrogen levels in the neonate: a comparison of two methods for sample collection.

We have compared a nasopharyngeal catheter method for breath sampling and a valved collection device. Sample quality was assessed by simultaneous oxygen measurement and reproducibility was checked by the analysis of 50 pairs of samples from four premature neonates. Both collection methods produced samples of highly variable quality suggesting a variable mixture of alveolar and non-alveolar air as well as differences in expired oxygen levels between babies. The mean oxygen levels were similar for both sampling techniques. Linear regression analysis of paired hydrogen results showed a highly significant correlation coefficient r = 0.73. This was improved markedly to r = 0.94 by normalization of hydrogen values based on observed oxygen levels, and so supports the earlier article by Robb and Davidson [1]. All breath hydrogen analyses require a measure of sample quality. Reproducible results and meaningful changes in hydrogen concentration in breath samples can only be achieved by correction according to sample quality. Correction to a common oxygen value should allow quantitative comparisons between patients. Samples were best collected from neonates by the nasopharyngeal catheter method, which least disturbed the patient, allowed multiple sample collection and gave lower oxygen and higher hydrogen values where large differences between pairs occurred.