Effects in premature infants of normalizing breath H2 concentrations with CO2: increased H2 concentration and reduced interaliquot variation.

Interval sampling of expired breath samples from the anterior nares is a feasible and noninvasive method for detecting elevated breath H2 and thus colonic carbohydrate fermentation, especially in nontherapeutic research studies of healthy premature infants. However, there may be a risk of falsely low breath H2 concentration and an unacceptable experimental error due to contamination with room air as well as with dead space air. We studied ten premature infants (28-32 week gestational age and 2-4 weeks postnatal age) who were receiving either a proprietary formula containing equal proportions of lactose and glucose polymer, or a similar formula in which lactose was the sole carbohydrate. In 70 breath samples (obtained 30-180 min after feeding) we assessed the coefficient of variation in breath hydrogen concentration among three aliquots obtained over a 3-5 min interval. Breath was collected from the anterior nares. The interaliquot coefficient of variation averaged 11% when expressed as parts per million per 5% CO2, compared to 19% when expressed as parts per million (Wilcoxon, p less than 0.001). Mean breath hydrogen concentration at each time period using the former method (parts per million per 5% CO2) was about 100% higher than when using the latter method. Although ventilation rate can alter alveolar CO2, normalizing for CO2 concentration reduces a major source of experimental error.