This study suggests that recent changes in food intake by infants with bronchiolitis closely reflect their oxygenation status. The 24h FI cutoff most strongly associated with hypoxia (odds ratio
13.8 [4.3 to 44.1]) was 50% of the usual amount, and this cutoff also had high negative predictive value. In the study population, there were 14 infants who took less than 50% of their usual food intake, and only one of them was not hypoxic. This 50% cutoff also represented the best compromise (moderate sensitivity, high specificity) for identifying infants with SpO2
95%. In addition, multivariate analysis showed that this parameter remained correlated with SpO2
95%, even after adjustment for age and intercostal retractions, while the other clinical and biological parameters showed no significant correlation with SpO2. This cutoff (50%) would have two advantages. First, if food intake during the past 24 hours remains more than half the usual amount, the likelihood of normal oxygenation status is 96% (or at least 91%, the lower limit of the confidence interval), indicating that bronchiolar obstruction is well tolerated and that pulse oximetry is probably unnecessary in the absence of associated gravity signs. In contrast, food intake below 50% of normal calls for pulse oximetry and medical attention.
This 50% cutoff has a certain tolerance: if 24h FI is 40%, then the negative predictive value only falls from 96% to 94% and the odds ratio from 13.8 to 12.2. The clinical implications therefore remain valid.
24h FI might be used as a screen to further testing. This might be useful for telephone triage or other settings where the history can be used to guide further evaluation, as monitoring a bronchiolitic child in the home.
Food intake has already been included in severity scores and guidelines, but only as “reduced or normal” [2
], or “good, reduced or poor” [8
]. The November 2006 Scottish guideline [11
] mentioned a
50% reduction in fluid intake in the previous 24 hours as a gravity sign, but this was based on expert opinion and not on published research.
Physiological studies of milk intake by infants with bronchiolitis [12
] have shown that the number of sucking and swallowing actions is the same as in healthy infants, but that milk flow during these two movements is reduced. In addition, swallowing movements are followed less by expiration and more by inspiration, and periods of apnea are twice as frequent after swallowing. Suckling is a muscular effort that must be coordinated with breathing, and this is more difficult when breathing is restricted by airway obstruction, when the child is tired, and when the muscles lack oxygen [13
Parents are generally aware of the amount of food their infant usually consumes, and can accurately recall food intake in the previous 24 hours, based on the amount prepared and the amount left, especially when their child is sick. Indeed, parents prepare the same quantity of milk (powder scoops and bottle graduations) each day for a given meal. The amount left in the bottle by healthy infants does not vary much, but increases during bronchiolitis. Thus, in practice, 24-hour percentage food intake is easy to calculate.
In theory, SpO2 would be an accurate marker of hypoxia, which is directly related through the hemoglobin dissociation curve. Hypoxia is the main complication of bronchiolitis. However, SpO2 is tricky to measure, requiring a considerable period of calm, and a correctly recorded arterial pulse. Correct sensor positioning may take several attempts. In addition, pediatric sensors are fragile, difficult to disinfect correctly, and expensive for single use. In the range of values generally measured in this setting (85-100%), a small change in SpO2 is associated with a large change in PaO2. The result is also observer-dependent and may vary over a 30-minute interval [14
]. This is why we verified all values below
95% by a second measurement, and only used the highest value.
The 95% SpO2 cutoff is commonly used to define hypoxia, and oxygen therapy is recommended if SpO2 is
90% for a prolonged period in an infant not at risk (term birth, no heart or lung disease and age more than 3 months) [5
Among the three types of retraction (suprasternal, intercostal and subcostal), intercostal retractions were most closely associated with SpO2
95%. Wang [13
] reported that the appreciation of retractions is observer-dependent, even when the observer is a healthcare professional, and that the appreciation may vary between two separate observations made 10 to 30 minutes apart. The degree of retractions is difficult to quantify, especially in borderline situations. In addition, they are difficult for parents to assess without specific training. Retractions are included in most scores, but these are not always easy to use in daily practice. Measurement of 24 hours percentage food intake seems to be far simpler for parents.
Hospitalization was associated with hypoxia and with age
2 months (French guidelines recommend hospitalization of all bronchiolitic infants less than 6 weeks old). Hypoxia was associated with 24h FI
50% and intercostal retractions. In the physician’s office, if SpO2 cannot be measured, these two signs, especially 24h FI
50%, in addition to age, can help to decide whether hospitalization is necessary.
This study may have certain limitations. First, the same pediatrician assessed all the data (24h FI, clinical signs of respiratory impairment, oxymetry) and took the decision regarding hospitalization. We tried to limit the subjective element by establishing the order of data collection and applying precautions during oxymetry.
Second, these results are applicable to patients presenting to community practices. Further studies of sicker infants (emergency departments, hospital inpatients) would be necessary to determine whether 24h FI remains valid in this population. It should be noted that most studies of severe adverse outcomes in this setting have involved inpatients, which limits their relevance to other situations [5
]. Finally, we excluded breast-feeding infants, as the amount of milk ingested could not easily be estimated, notably from the suckling time [15