High Flow Nasal Cannula and Poor Outcomes? Response to Taha, et al (2016)

Taha DK, Kornhauser M, Greenspan JS, Dysart KC & Aghai ZH. High Flow Nasal Cannula Use Is Associated with Increased Morbidity and Length of Hospitalization in Extremely Low Birth Weight Infants. The Journal of Pediatrics, 2016. Volume 173 , 50 – 55.e1.

See also:
Roberts, Calum T.Owen, Louise S.Yoder, Bradley A.Manley, Brett J. et al. Chicken or egg? Dangers in the interpretation of retrospective studies. The Journal of Pediatrics, 2016. Volume 178 , 309.
Bromiker, Ruben, Kugelman, Amir et al. High flow nasal cannula and poor outcomes? The Journal of Pediatrics, 2016. Volume 178, 308.

Summary of the Work
Taha and colleagues reported on a retrospective database analysis (Alere Neonatal Database) comparing 2487 extremely low birth weight infants (ELBW; ≤ 1000 g at birth) who received CPAP and/or HFNC therapy during their NICU stay. The groups for this analysis were CPAP-only (n = 941), HFNC-only (n = 333) and HFNC with or without CPAP (HFNC ± CPAP; n = 1546, where 78% did receive CPAP as well). Primary outcomes were BPD and mortality, where HFNC-only compared to CPAP-only was associated with greater BPD (P < 0.05; 56.8% vs 50.4%) and mortality (P < 0.05; 52.2% vs 42.2%). The analyses predict that the odds for developing BPD/mortality was greater for HFNC ± CPAP (OR 1.085; p < 0.01) than for CPAP-only.

While the association between HFNC and BPD/mortality should raise concern, there are major limitations of this analysis. Please consider the following points when interpreting this study:

The primary independent variable, HFNC, was not defined.
The current paper does not define HFNC (e.g., by what minimal liter flow), nor do they report what the Alere Neonatal Database considers to be HFNC when the data are entered from each center. The authors state that the database does not record liter flow (HFNC) or pressure (CPAP), nor does the database record the order in which the therapies were used.

It is well known that clinical practice varies among neonatal units with regard to the use of HFNC, and there is controversy over mechanisms of action, appropriate flow ranges and which infants are best served. HFNC is so heterogeneously applied that manufacturers market opposing cannulae design and flow characteristics, flow circuitry and recommendations for bulk flow use. In the current paper, the authors point out that acute outcomes are substantially (and significantly) different between studies that used low flow rates (≤2 L/min; Campbell et al) versus higher flow rates (8 L/min; Collins et al). A pooled analysis of the data from the numerous recent acute randomized, controlled trials of HFNC in the NICU (Rotta et al, 2015) shows a significant difference in re-intubation rates as a function of device platform. Therefore, a preponderance of the evidence dictates that outcomes from HFNC would depend on how it is applied, and that is not controlled for in this analysis.

The HFNC group had a longer duration of mechanical ventilation
Median ventilatory days for the HFNC-only group was 39% greater than for CPAP-only group (25 d vs 18 d; note, 30 d for the HFNC ± CPAP group). These data alone would suggest that the HFNC infants were of higher acuity and experienced more ventilator-induced lung injury independent of non-invasive ventilatory strategy implementation. Both ventilator-induced lung injury and acuity are well known to be associated with both BPD and death. To this point, the following are excerpts from the authors’ introduction, which were not subsequently addressed in the Discussion section:

  • “contributing factors to the development of BPD…ventilator-induced lung injury”
  • “In an effort to reduce injury from mechanical ventilation, noninvasive modes of respiratory support are increasingly being used.”
  • “The use of CPAP may reduce BPD as it minimizes duration of mechanical ventilation.”

The increase in mechanical ventilation for HFNC groups was not likely the result of HFNC use given that the variable ‘mechanical ventilation initiated on day 1’ ranged from 70% to 78% of the infants across groups. The percent of infants receiving multiple courses of ventilation were similar between HFNC-only and CPAP-only (53% vs 51%, respectively).

Other indications that the HFNC infants may have been of higher acuity were the greater use of postnatal steroids, the overall longer hospital length of stay, and the trend where the CPAP-only group had the lowest percentage of infants with a 5-min Apgar score <5.

In the HFNC ± CPAP group, 78% of infants did receive CPAP
The current paper compares HFNC-only and HFNC ± CPAP groups to CPAP-only, and the finding for a significant odds ratio of BPD / death was associated with the HFNC ± CPAP group. However, when considering results related to the HFNC ± CPAP group, it must be noted that 78% of these infants did receive CPAP. Therefore, this group cannot be inconspicuously compared to CPAP; they were CPAP. Moreover, as stated above, there is no definition as to when and why HFNC was initiated over CPAP. For example, in some cases CPAP infants could have been transferred to HFNC because of skin breakdown from CPAP overuse. HFNC could have been initiated as a stepdown from CPAP for sicker infants for whom a jump to simple oxygen was not considered clinically reasonable. Hence, this sample is likely biased, which would explain the consistency in having the poorest outcomes.

Nonetheless, the HFNC without CPAP infants are the same infants that make up the HFNC-only group, therefore the rise in percentages of both BPD and death from the HFNC-only group to the HFNC ± CPAP group must be associated with the infants in that group who received CPAP.

The sub-analyses presented by the authors are in disagreement on the primary outcomes, and no reference was made to the McQueen Paper
This study looked at infants from 466 level II and Level III NICUs, where 39% of the infants came from 378 low-volume centers, each of which contributed less than 10 infants to this analysis. A major confounding factor in outcomes would be related to practice patterns among this wide array of centers. Looking at data from just the 378 low-volume centers (n = 976 infants), there was no difference between groups in BPD or death despite similar differences in the other outcome variables. This would indicate that practice patterns in the remaining 88 high-volume centers (n = 1511) could have contributed to the overall differences in these primary outcomes. Nonetheless, analyses of the subsets of data did not agree with respect to BPD and morbidity.

A strikingly similar database study was recently published (McQueen et al, 2015) comparing respiratory outcomes with HFNC using the VLBW Vermont Oxford Network Database (n = 177,962 infants). In the McQueen study, lesser BPD (oxygen use at 36 weeks) was found in the HFNC cohort, with no differences in other outcomes or adverse events. McQueen and colleagues did a much better job of defining HFNC use based on the published mechanistic work in this field and qualifying the use of HFNC in the centers representing these data. The current authors were remiss to not mention this parallel study, and hence made no attempt to explain the disparity between the outcomes.

[1] Rotta, A; Speicher, R; Shein, S; Speicher, D. HIGH FLOW NASAL CANNULA THERAPY IN PRETERM INFANTS A POOLED ANALYSIS. Critical Care Medicine: December 2014 – Volume 42 – Issue 12 – p A1541.
[2] McQueen M, Rojas J, Sun Shyan, Tero R, Ives K, Bednarek F, Owens L, Dysart K, Dungan G, Shaffer T, Miller T. Safety and long term outcomes with high flow nasal cannula therapy in neonatology: a large retrospective cohort study. J Pulm Respir Med. 2014 Dec; 4(6): 216.
[3] Collins C, Holberton J, Barfield C, Davis P. A randomized controlled trial to compare heated humidified high-flow nasal cannulae with nasal continuous positive airway pressure postextubation in premature infants. J Pediatrics. 2013 May; 162: 949-54.
[4] D M Campbell, P S Shah, V Shah & E N Kelly. Nasal continuous positive airway pressure from high flow cannula versus Infant Flow for preterm infants. Journal of Perinatology (2006) 26, 546–549. doi:10.1038/sj.jp.7211561.

This webpage contains links to third party abstracts and/or publications. With respect to those materials, please note that Vapotherm’s Hi-VNI® technology is a tool for treating the signs and symptoms of respiratory distress in patients for whom prescribers desire to add heat and moisture to breathing gases. The linked materials may describe certain outcomes in relation to the use of Vapotherm’s Hi-VNI Technology, but individual results may vary. Practitioners should refer to the full indications for use and operating instructions of any products referenced herein before prescribing them.