OBJECTIVE. The goal was to estimate the level of delivered continuous positive airway pressure by measuring oral cavity pressure with the mouth closed in infants of various weights and ages treated with heated, humidified high-flow nasal cannula at flow rates of 1–5 L/minute. We hypothesized that clinically relevant levels of continuous positive airway pressure would not be achieved if a nasal leak is maintained.
METHODS. After performing bench measurements and demonstrating that oral cavity pressure closely approximated levels of traditionally applied nasal continuous positive airway pressure, we successfully measured oral cavity pressure during heated, humidified, high-flow nasal cannula treatment in 27 infants. Small (outer diameter: 0.2 cm) cannulae were used for all infants, and flow rates were left as ordered by providers.
RESULTS. Bench measurements showed that, for any given leak size, there was a nearly linear relationship between flow rate and pressure. The highest pressure achieved was 4.5 cmH2O (flow rate: 8 L/minute; leak: 3 mm). In our study infants (postmenstrual age: 29.1–44.7 weeks; weight: 835–3735 g; flow rate: 1–5 L/minute), no pressure was generated with the mouth open at any flow rate. With the mouth closed, the oral cavity pressure was related to both flow rate and weight. For infants of ≤1500 g, there was a linear relationship between flow rate and oral cavity pressure.
CONCLUSIONS. Oral cavity pressure can estimate the level of continuous positive airway pressure. Continuous positive airway pressure generated with heated, humidified, high-flow nasal cannula treatment depends on the flow rate and weight. Only in the smallest infants with the highest flow rates, with the mouth fully closed, can clinically significant but unpredictable levels of continuous positive airway pressure be achieved. We conclude that heated, humidified high-flow nasal cannula should not be used as a replacement for delivering continuous positive airway pressure.
Nasal continuous positive airway pressure (NCPAP) therapy is used as a primary support modality for infants with respiratory distress syndrome and apnea of prematurity, as well as for postextubation respiratory support. The benefits of continuous positive airway pressure (CPAP) therapy derive from alveolar recruitment and reduction of airway collapse. The pulmonary physiologic effects include increased transpulmonary pressure, functional residual capacity, and compliance, decreased pulmonary resistance, and splinting of the airway and diaphragm.1,2 After weaning from CPAP therapy, oxygen is often delivered through nasal cannulae at low flow rates (up to 0.5 L/minute). This method is frequently preferred by caregivers, compared with other methods of supplemental oxygen delivery, because of the ease of administration and the ability to care for the infant without limiting interactions with the environment. It is also thought not to alter respiratory mechanics.3
In the past few years, higher nasal cannula flow rates (≥1 L/minute) have been used increasingly in many ICUs, with the perception that this form of therapy can provide many of the same benefits as conventional NCPAP therapy. Moreover, there is some evidence that providing higher flow rates via nasal cannulae can produce some level of CPAP in premature neonates.4 However, the use of this technique with high flow rates may generate excessive CPAP, especially when the mouth is closed and tightly fitting nasal cannulae are used. In one study of premature infants, more tightly fitting nasal cannulae (outer diameter [OD]: 0.3 cm) generated positive end distending pressures at flow rates of 1 and 2 L/minute, whereas more loosely fitting nasal cannulae (OD: 0.2 cm) did not.5
Unlike conventional NCPAP therapy, in which the infant breathes from a pressurized circuit, flow through nasal cannulae is directed entirely into the nasopharynx, with the only escape routes being the mouth and the nose. With most NCPAP devices, pressure in the circuit is controlled with a valve that provides an “escape route” in addition to the nose and mouth. Therefore, with the mouth closed and with snugly fitting nasal prongs, the pressure in the oropharynx is controllable within the range selected by the health care provider. With nasal cannulae, however, and with the mouth closed, pressure in the oropharynx depends entirely on the flow and the snugness of the fit of the nasal cannulae and therefore is not directly controllable. Moreover, some NCPAP devices are designed to maintain pressure even when the mouth is open; with nasal cannula flow systems, virtually all of the flow escapes through the mouth when it is open.1 In addition, increasing flow rates with conventional nasal cannulae, in which the gases are not heated to body temperature and are not fully saturated with water, may increase heat loss and produce nasal drying effects, leading to skin breakdown and increasing the risk for infection.6
Vapotherm (Stevensville, MD) and Fisher & Paykel Healthcare (Irvine, CA) manufacture high-flow nasal cannula devices that deliver breathing gases heated to near body temperature and highly saturated with water vapor, through small nasal cannulae at relatively high flow rates (1–8 L/minute), that is, heated, humidified, high-flow, nasal cannula (HHHFNC) devices.7,8 In addition to providing heated and humidified breathing gases, HHHFNC devices may provide an additional benefit by generating clinically relevant levels of CPAP. Despite its increasing popularity and the potential of providing benefits similar to those of conventional NCPAP therapy, this technique has not been well studied, and results of published reports regarding safety have been conflicting. Flow rates of 2 to 2.5 L/minute through nasal cannulae have been reported to produce both excessively high5 and safe and clinically relevant levels of CPAP.4 Therefore, we chose to use only the smallest (OD: 0.2 cm) nasal cannulae for our study, on the basis of evidence that larger cannulae can potentially generate unsafe levels of CPAP.5 At the time of the study, no infants were treated with flow rates of >5 L/minute, and the flow rates ordered for individual patients were at the discretion of the health care providers. We hypothesized that the amount of CPAP delivered with HHHFNC systems with the mouth closed would depend on the rate of flow and the size of the leak around the nasal cannula. We also hypothesized that, with the smallest (OD: 0.2 cm) nasal cannulae, HHHFNC devices would not generate clinically relevant levels of CPAP, except perhaps for the very smallest infants.
The study was conducted in the intensive care nursery at the Dartmouth-Hitchcock Medical Center between January 2005 and April 2006 and was approved with both Vapotherm 2000i and Fisher & Paykel devices by the Dartmouth College/Medical School Committee for the Protection of Human Subjects. Informed parental consent was obtained for each infant in the study. The Vapotherm system was used until December 2005; beginning in January, 2006, the Fisher & Paykel device was used. The decision to introduce the latter system was made on the basis of the cost analysis, ease of sterilization, and reports issued by the Centers for Disease Control and Prevention regarding possible Vapotherm system contamination with Ralstonia sp. There were no cases of Vapotherm-related infections in our nursery; however, this potential safety consideration expedited our decision to replace Vapotherm devices with Fisher & Paykel devices.
The Vapotherm and Fisher & Paykel devices both use a high-flow closed system and, at the time of the study, were approved by the Food and Drug Administration for pediatric use. The smallest (OD: 0.2 cm) nasal cannulae (DTPV9007; Vapotherm) were used with both systems in the current study. All infants treated with HHHFNC therapy were potential candidates for enrollment. Exclusion criteria included neuromuscular disorders, multiple congenital and/or chromosomal anomalies, and severe neurologic impairment, including grade 3 or 4 intraventricular hemorrhage.
Before human measurements were performed, bench data were obtained to estimate the pressure levels that might be encountered in the study. Vapotherm neonatal cannulae (OD: 0.2 cm) were inserted snugly into the port of an anesthesia bag, which was used to simulate an infant's lung. The anesthesia bag was initially filled with gas at a pressure of 20 cmH2O, to verify the absence of leaks around the cannula. Another port of the anesthesia bag was configured to allow 3 opening sizes (3.0, 5.0, and 8.0 mm). The sizes were chosen to cover the range of leaks that might be encountered around the nasal cannula with infants of different sizes. Three pressure measurements were made for each opening size at each flow rate (1.0–8.0 L/minute for the Vapotherm device and 1–4 L/minute for the Fisher & Paykel device [a range approved by the Food and Drug Administration for this device at the time of the study], in 1.0-L/minute increments), by using a calibrated pressure transducer.
Oral Cavity Pressure Measurements as an Estimate of Delivered CPAP During HHHFNC Treatment
All infants treated with high-flow nasal therapy with a Vapotherm or Fisher & Paykel device were studied without sedation, in the supine position and during quiet sleep (quiet, eyes closed, regular breathing, and no rapid eye movements or twitching). Each infant was studied at the flow rate and oxygen fraction that had been ordered by the provider, which were not altered during the study. After calibration of the transducer (model CD15; Validyne Engineering, Northridge, CA), an 8-French feeding catheter was inserted ∼3 cm into the oral cavity and connected to the transducer. After confirmation that the infant was in quiet sleep, baseline measurements were recorded for 1 to 2 minutes, with the infant's mouth open. After completion of the control measurements, the infant's mouth was closed with a gentle chin lift and was held closed for 1 to 2 minutes while pressure was recorded continuously. The feeding catheters were not left in place for >15 minutes. The measurements were repeated 3 times on 1 day for each infant, and respiratory rate, oxyhemoglobin saturation, and heart rate were monitored continuously. At the time of the study, the Vapotherm device was approved for a maximal flow rate of 8 L/minute and the Fisher & Paykel device was approved for a maximum of 4 L/minute. It was our practice not to exceed 5 L/minute for any infant receiving HHHFNC therapy.
Oral Cavity Pressure Measurements With Standard “Bubble” NCPAP Treatment
To validate the measurement technique and to confirm that oral cavity pressure measurements could closely estimate the level of clinically ordered NCPAP, mouth pressure was measured for 5 infants receiving standard bubble NCPAP therapy. The measurement technique was similar to that used for measurements with the HHHFNC devices, using an 8-French feeding tube with the mouth held closed. Infants were in quiet sleep, and no alterations in the infant's clinically ordered CPAP level or fraction of inspired oxygen were made before the measurements were performed. For each infant, 3 measurements (1–2 minutes) were made and compared with baseline measurements made with the mouth open.
Data Reduction and Statistical Analyses
All data were digitized at 50 Hz and recorded continuously (DI-194RS; Data Translation Software, Akron, OH). For each infant, the 3 measurements were averaged. To determine whether mouth pressure was related to flow rate and nasal leakage, a multivariate regression analysis was performed by using the weight of the infant, as an index for nasal orifice size, and flow rate as independent variables and the mean or maximal pressure generated as the dependent variable. In addition, data for infants who weighed ≤1500 g (minimal nasal leakage) at the time of the study were analyzed separately, to determine whether flow rate could predict delivered CPAP for smaller infants.
At the highest flow rate (8.0 L/minute) and the smallest opening (3 mm), the highest level of CPAP achieved with the Vapotherm device was 4.5 cmH2O. With larger opening sizes, considerably lower pressures were achieved, that is, 2.1 cmH2O with an opening size of 5.0 mm and 0.5 cmH2O with an opening size of 8 mm. The results obtained with the Fisher & Paykel device generally paralleled those obtained with the Vapotherm system. At a flow rate of 5.0 L/minute and the smallest opening size (3 mm), the highest pressure achieved was 1.6 cmH2O. When the opening size was increased to 8 mm, only 0.08 cmH2O was generated (Fig 1).
Pressure Measurements During Standard Bubble NCPAP Treatment
Measurements made during standard bubble NCPAP therapy with the mouth closed for 5 infants closely approximated the set NCPAP pressures. An example, showing oral cavity pressures with set NCPAP values of 4 and 6 cmH2O, is shown in Fig 2.
Oral Cavity Pressure Measurements During HHHFNC Treatment
Twenty-seven infants were enrolled in the study. Sixteen were studied by using the Vapotherm device and 11 by using the Fisher & Paykel device. At the time of the study, the infants ranged in postmenstrual age from 29.1 to 44.7 weeks, weighed 835 to 3735 g, and were treated with HHHFNC flow rates of 1 to 5 L/minute. The demographic data are summarized in Table 1.
With the infant's mouth open, HHHFNC treatment generated no increase in pressure for any infant, at any flow rate. The highest oral cavity pressures of 4.3 to 4.8 cmH2O were recorded for infants weighing <1500 g (range: 900–1470 g), at a flow rate of 4.0 L/minute. In general, in this group of small infants, flow rates of 1 to 3 L/minute produced mouth pressures of 0.6 to 4.1 cmH2O. The highest pressure of 4.8 cmH2O was achieved with a flow rate of 4 L/minute in an infant who weighed 1281 g. In contrast, for infants weighing >1500 g, in general, no pressures were generated >2.6 cmH2O, even with the highest flow rate of 5.0 L/minute, except 1 infant (attributed to difficult recording). For infants weighing <1500 g, there was a linear relationship between flow rate and maximal oral cavity pressure. Maximal pressure = 1.32 × flow − 0.86; r2 = 0.772; P = 000077. The units for flow are L/minute. For infants weighing >1500 g, no relationship was evident (Fig 3). For all infants, the maximal oral cavity pressure generated was related to both flow rate and weight in a multivariate regression analysis. Maximal pressure = (0.444 × flow) − (0.001 × weight + 3.022); r2 = 0.701; P = .0004; weight, P = .00008; flow, P = .01746 (Fig 4). The units for flow are L/minute and grams for weight.
High-flow nasal cannula devices have been used increasingly in many NICUs (up to 64% increase in one report9) and apparently have been accepted rapidly because of their ease of use and their ability to provide heated, humidified, oxygen/air gas mixtures.3 There have been numerous reports of their use to treat apnea of prematurity, to prevent reintubation, and in some cases to replace conventional NCPAP therapy. A recent retrospective study that examined the increased use of HHHFNC treatment replacing NCPAP therapy for infants at <30 weeks of gestation with respiratory disease, over a 3-year period, showed no differences in outcomes.9
Our study was performed to determine whether clinically significant levels of CPAP could be achieved with HHHFNC treatment by using flow rates up to 5 L/minute and to gain insight regarding the factors responsible. We hypothesized that the level of CPAP (estimated as oral cavity pressure) generated with HHHFNC treatment would be related to the flow rate and to the size of the nasal leak, using body weight as an index. Flow rates in our study were limited to 5 L/minute and were based on provider decisions. At the time of the study, flow rates of up to 8 L/minute were approved for neonates for the Vapotherm device and up to 4 L/minute for the Fisher & Paykel device. Of note, the current Fisher & Paykel product (RT329 infant oxygen therapy) has been designed and currently is being marketed to deliver flow rates of 0.3 to 8 L/minute in the pediatric population.
The main finding of our study was that HHHFNC therapy using the smallest nasal cannula (OD: 0.2 cm) could achieve a clinically relevant level of CPAP only in the very smallest infants. We also showed that measurements of oral cavity pressure closely estimated the clinically prescribed CPAP level. Importantly, CPAP with HHHFNC devices was achieved only with the mouth closed. With the mouth open, no pressure was generated in the oral cavity at any flow rate. For infants weighing >1500 g, clinically relevant levels of CPAP could not be generated even with the mouth fully closed, most likely because of the presence of larger nasal leaks. When all infants were included in the analysis, the oral cavity pressure generated with HHHFNC treatment was related to both flow rate and weight.
On the basis of our findings, we speculate that the relationship with body weight is attributable to the related size of the leak around the nasal cannula. The relationship between body weight and nares size may not be linear. The size of the nares may differ in infants of the same weight because of variability in anatomic features, and the relationship between nares size and weight may change over time for the same infant after prolonged use of NCPAP treatment or nasal intubation. Nevertheless, we used this assumption as a first approximation because currently there is no practical, simple method to measure nares size or the degree of nasal leakage around nasal cannulae objectively.
Our analysis was performed with a relatively small number of infants, and the decision to analyze infants of <1500 g separately was made posthoc, to separate out the smallest infants. Nevertheless, the relationship between flow rate and generated oral cavity pressure for the 13 infants of ≤1500 g was highly significant, with a probability of 8 of 100000 that this occurred by chance alone. The power calculated by using an α of .05, the SDs of the independent variable (flow rate) and the regression errors, a slope of 1.32, and 13 subjects, assuming that we did not assign the flow rates, was 0.99. The infants we studied were heterogeneous with respect to clinical diagnoses, including chronic lung disease, apnea of prematurity, periventricular leukomalacia, retinopathy of prematurity, and necrotizing enterocolitis. Our data illustrate only the factors that determine the level of CPAP that can be generated, and we make no predictions regarding the clinical usefulness of this therapy in treating conditions such as apnea of prematurity.
Our results and those of others raise issues of safety related to HHHFNC therapy, particularly when larger (presumably more tightly fitting) cannulae are used at high flow rates for smaller infants and the mouth leakage is minimized. It is important to emphasize that conventional NCPAP devices have pressure-limiting capabilities, which do not exist with HHHFNC systems.
All of our measurements were performed by using the smallest available (OD: 0.2 cm) nasal cannulae, which allowed some degree of nasal leakage. Particular caution is warranted if a larger nasal cannula is used. We speculate that, if the nasal leak is eliminated with the use of a cannula that obstructs the nares completely, then dangerously high levels of distending pressure could be generated during periods when the mouth is closed, because there is no effective “pop-off” available. It is also important to emphasize that only 2 infants of <1000 g were studied, one with a flow rate of 3 L/minute and the other with a flow rate of 4 L/minute, which generated oral cavity pressures of 4.38 and 4.14 cmH2O, respectively. Presumably, flow rates of 4 L/minute in infants of lower weight might generate even higher (and perhaps dangerous) levels of CPAP. Moreover, both Vapotherm and Fisher & Paykel devices are currently approved for flow rates up to 8 L/minute. On the basis of our limited data and our bench studies, the use of these devices at flow rates of >5 L/minute, even with the smallest nasal cannulae, for infants of ≤1000 g may generate dangerously high levels of CPAP and cannot be recommended. Clearly, these speculations warrant additional studies.
It has been proposed that high-flow nasal cannula systems can provide therapeutic benefits and replace conventional NCPAP therapy, but few studies have reported the ability of these devices to generate a consistent level of CPAP. Moreover, not all studies used systems with heated and humidified breathing gases, and varying methods have been used to estimate distending pressures. Few studies have controlled for the size of leaks around the nasal cannulae or the degree of mouth opening. The use of esophageal pressure measurements to estimate distending pressures generated in premature infants has yielded conflicting results, with one study reporting that the equivalent of 9.8 cmH2O could be generated with only 2 L/minute of flow by using a nasal cannula of 0.3-cm OD (heating and humidification not reported)5 and another reporting that clinically relevant levels of CPAP could be generated with flow rates of up to 2.5 L/minute (nonheated Hudson humidification).4 The authors of the second study suggested that the flow rate needed to generate an adequate distending pressure could be estimated with the following equation: flow rate (in liters per minute) = 0.92 + (0.68 × weight [in kilograms]). In the first study, significant pressure was not generated with smaller (OD: 0.2 cm) cannulae; in the second study, the size of the nasal cannulae was not reported. Therefore, the differences in the results could have reflected differences in cannula size.
A recent, small, randomized, controlled trial compared the feasibility of using high-flow nasal cannula treatment to replace conventional NCPAP treatment for prevention of reintubation among preterm infants with a birth weight of ≤1250 g.10 In that study, a nonheated bubble humidifier system was used with “standard” nasal cannulae of an unstated size. The authors used the aforementioned formula4 to calculate the flow rate needed to generate the clinically relevant CPAP level, but they did not measure the pressures generated. It was concluded that “CPAP” delivered with high-flow nasal cannulae failed to maintain extubation status among preterm infants of ≤1250 g as effectively as did NCPAP therapy.10 Interpretation of the results of that study is difficult because of the small sample size, the unknown size of the nasal cannulae, and the lack of information about the actual pressures generated.
Another recent study, describing the pulmonary mechanics in preterm neonates during HHHFNC treatment (Vapotherm device; flow rates of 3, 4, and 5 L/minute; unknown cannula size) and NCPAP therapy (6 cmH2O), found no differences in the work of breathing at any flow rate.11 Interestingly, there was a significant increase in end distending pressure from baseline (no HHHFNC or NCPAP treatment) with a HHHFNC flow rate of 5 L/minute but not with lower HHHFNC flow rates or with NCPAP of 6 cmH2O.11 The reasons for these findings are unclear, but they suggest that measurements of esophageal pressures may not be useful for estimating the level of CPAP that may be generated with HHHFNC treatment.
Our technique for estimating the level of CPAP by measuring oral cavity pressure is effective, simple, and relatively noninvasive and may provide an alternative to making more-invasive measurements of esophageal balloon pressure. We occasionally encountered artifacts attributable to swallowing, movement, or secretions in the measuring catheters, which made some recordings uninterpretable. However, good-quality, reproducible recordings were obtained easily when the infants were in quiet sleep.
There are other benefits of delivering heated humidified oxygen that are unrelated to potential changes in pulmonary mechanics. Both the Vapotherm and Fisher & Paykel devices overcome the disadvantages of commonly used nasal cannula therapy, in which the delivered gas is well below body temperature and only partially humidified. Medical gases have little water content at room temperature, and their delivery to the airways without adequate humidification and warming may lead to airway dysfunction and negative respiratory outcomes. Cooling and loss of water from the airways may impair mucociliary transport, increase fluid osmolality, promote bronchospasm, and increase the viscosity of airway secretions. Moreover, considerable energy is required to heat and to humidify gas delivered into the nose, potentially interfering with optimal nutrition and growth.7,12 On the basis of studies with very low birth weight infants undergoing ventilation, the delivery of nonhumidified gas may lead to increases in air leaks, more-severe chronic lung disease, impaired surfactant activity, and changes in pulmonary mechanics.13,14 Delivery of cold, relatively dry gases, particularly at high flow rates, to the nose may also lead to nasal mucosal injury and bleeding, resulting in pain and creating a portal of entry for infectious agents.6 One could argue that these devices, used at low flow rates (<1 L/minute), should replace standard, nonheated, partially humidified, nasal cannula systems in use in most NICUs. Considering the emerging clinical usage of high-flow nasal cannula therapy, there is a need for studies that address this question.
Both the Vapotherm and Fisher & Paykel device use the process of vaporization, which generates a molecular distribution of water that is free of droplet water and provides vapor that is nearly 100% humidified at body temperature (Vapotherm: 99. 9% relative humidity at 37°C; Fisher & Paykel: 82%) and theoretically is unable to carry infectious agents.7,8,12 Particular attention, however, needs to be paid to the disinfection and sterilization processes for any humidification device, because microbial colonization of the delivery system remains a possibility.12,15,16
We conclude that HHHFNC therapy using cannulae of 0.2-cm OD can generate some level of CPAP when the mouth is closed. On the basis of our results, the pressure generated is more predictable for smaller preterm infants. In general, the amount of pressure generated is related to the flow rate, the size of the leak around the nasal cannula, and the degree of mouth opening. In this regard, measurement of oral cavity pressure is a relatively simple, noninvasive method of estimating the level of CPAP generated with HHHFNC treatment. Our results and those of others also raise important safety and monitoring issues for the use of these devices. Particular caution is warranted if larger nasal cannulae are used. We speculate that, if the nasal leak is eliminated with the use of cannulae that obstruct the nares completely, then dangerously high distending pressures may be generated when the mouth is closed. On the basis of our findings, we suggest that high-flow nasal cannula therapy should not be used as a routine replacement for CPAP therapy. Future studies addressing the usefulness of high-flow nasal cannula therapy in treating apnea of prematurity, or as a substitute for conventional CPAP, should take into account the important relationships between flow rate and nasal leaks, the level of CPAP generated, and the degree of heating and humidification.
We thank all of the nurses and respiratory therapists who assisted with our work and, most of all, the families that participated in the study.
- Accepted July 27, 2007.
- Address correspondence to Zuzanna J. Kubicka, MD, Department of Pediatrics, Division of Neonatal Perinatal Medicine, Dartmouth-Hitchcock Medical Center, 1 Medical Center Dr, Lebanon, NH 03756. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Goldsmith JP, Karotkin EH. Continuous positive airway pressure. In: Goldsmith JP, Karotkin EH, eds. Assisted Ventilation of the Neonate. 4th ed. Philadelphia, PA: Saunders; 2003:127–147
- ↵Finer NN. Nasal cannula use in the preterm infant: oxygen or pressure? Pediatrics.2005;116 :1216– 1217
- ↵Sreenan C, Lemke RP, Hudson-Mason A, Osiovich H. High-flow nasal cannulae in the management of apnea of prematurity: a comparison with conventional nasal continuous positive airway pressure. Pediatrics.2001;107 :1081– 1083
- ↵Locke RG, Wolfson MR, Shaffer TH, Rubenstein SD, Greenspan JS. Inadvertent administration of positive end-distending pressure during nasal cannula flow. Pediatrics.1993;91 :135– 138
- Copyright © 2008 by the American Academy of Pediatrics