Kanthimathinathan and Kneyber Critical Care (2020) 24:116

https://doi.org/10.1186/s13054-020-2837-3

LETTER Open Access

Impact of HFOV in pARDS outcomes:

questions remain
Hari Krishnan Kanthimathinathan1* and Martin C. J. Kneyber2

We read with interest the analysis comparing outcomes analysis. Curiously, the odds ratios of mortality for
in paediatric acute respiratory distress syndrome mild, moderate and severe pARDS in the logistic
(pARDS) among those who required high-frequency os- regression analysis were 1.61, 1.02, and 1.00
cillatory ventilation (HFOV) as ‘rescue therapy’, com- respectively. This paradoxical trend may indicate an
pared to those who did not [1]. The authors must be interaction between the use of HFOV and presence
congratulated for using multiple matching techniques to of moderate (or) severe pARDS, rather than the risk
account for some confounders. We also commend them of HFOV per se. Was an interaction between the
for making the code used for analysis public [2]. We use of HFOV and severity of pARDS tested and
wish to clarify certain aspects of the study that may aid accounted for in the logistic regression analysis?
readers understand the context of the analysis better. 3. The intention to match patients in a propensity- or
genetic-matched study is to account for con-
1. The dataset used for genetic matching included at founders present, when an intervention was indi-
least 68 ‘matched-pairs’ of severe pARDS. However, cated, between those that received the intervention
the 68 ‘matches’ were derived from a pool of less compared to those that did not. The use of day 1
than 39 ‘real’ patients with severe pARDS in the OI, rather than the OI immediately prior to the
non-HFOV cohort. While multiple matching with initiation of HFOV, may have systematically under-
replacement may improve matching, it would be estimated the true severity of pARDS in the HFOV
helpful to understand how many multiple matches group, given that HFOV was used as a ‘rescue
were made from the smaller pool of non-HFOV pa- mode’ anytime within 7 days. An equivalent assign-
tients and whether statistical adjustments were ment of pARDS severity could be done using the
made for the duplicates in addition to a pairwise ‘highest OI’ within the first 7 days in the non-
comparison. HFOV group. Was the use of OI prior to initiation
2. The authors reported that the sensitivity analysis of HFOV, rather than day 1 OI, considered for
showed consistent findings with the primary analysis?

Authors’ response
Judith Ju Ming Wong, Siqi Liu, Mengling Feng, Jan Hau Lee

We thank Drs. Krishnan and Kneyber for their letter 1. In the genetic matching (GM) model, we used
and the opportunity to clarify questions that remain. matching with replacement aiming to have a larger
matched cohort [1]. Among the 67 uniquely
This comment refers to the article available at https://doi.org/10.1186/ matched non-high frequency oscillatory ventilation
s13054-020-2741-x. (HFOV) patients, majority were matched once [40/
* Correspondence: [email protected] 67 (59.7%)]. The remainder were matched twice
1
PICU, Birmingham Children’s Hospital, Steelhouse Lane, Birmingham B4
6NH, UK [16/67 (23.9%)], thrice [5/67 (7.5%)] and more than
Full list of author information is available at the end of the article

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if
changes were made. The images or other third party material in this article are included in the article's Creative Commons
licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons
licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain
permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the
data made available in this article, unless otherwise stated in a credit line to the data.
Kanthimathinathan and Kneyber Critical Care (2020) 24:116 Page 2 of 2

three times [6/67 (9.0%)]. No adjustments were Funding

made for multiple matched pairs. However, experi- None

ments of GM without replacement showed consist- Availability of data and materials

ent findings with our main analysis [odds ratio (OR) Not applicable
for mortality 1.9, 95% confidence interval (CI) 1.0
Ethics approval and consent to participate
to 4.0; p = 0.09] lending confidence to the results. Not applicable
2. We performed logistic regression with interaction
term between HFOV use and paediatric acute Consent for publication
Not applicable
respiratory distress syndrome (PARDS) severity and
compared this with the original model without the Competing interests
interaction term using chi-square test and the MCJK is a Principal Investigator of “PROSPECT” randomized control trial. HK’s
institution may participate in the trial.
‘ANOVA’ function in R software. There was no im-
provement (p = 0.16) with introducing the inter- Author details
1
action term, and hence, this was omitted. PICU, Birmingham Children’s Hospital, Steelhouse Lane, Birmingham B4
6NH, UK. 2Department of Paediatrics, Division of Paediatric Critical Care
We hypothesized that HFOV use is beneficial in Medicine, Beatrix Children’s Hospital, University Medical Center Groningen,
certain patient subgroups. We reported subgroup Huispost CA 80, P.O. Box 30.001, 9700 RB Groningen, The Netherlands.
analyses (severe and non-severe PARDS) in the sup-
Received: 28 February 2020 Accepted: 19 March 2020
plementary material (Table S3.2). A trend towards
harm was associated with HFOV use in the non-
severe [OR 2.9 (95% CI 1.1 to 8.1); p = 0.06] and References
1. Wong JJ-M, Liu S, Dang H, Anantasit N, Phan PH, Phumeetham S, et al. The
severe [OR 1.6 (95% CI 0.7 to 3.5); p = 0.29] cohort, impact of high frequency oscillatory ventilation on mortality in paediatric
but the effect size in severe PARDS seemed lower. acute respiratory distress syndrome. Crit Care. 2020;24(1):31.
This paradoxical trend was similar to the multivari- 2. Siqi L. R code for ‘The impact of high frequency oscillatory ventilation on
mortality in paediatric acute respiratory distress syndrome’ 2019. [cited 2020
ate analysis. However, this hypothesis may only be Feb 10]. Available from: https://github.com/nus-mornin-lab/KKH/.
answered with the completion of the PROSPect trial 3. Yehya N, Thomas NJ, Khemani RG. Risk stratification using oxygenation in
(NCT03896763). the first 24 hours of pediatric acute respiratory distress syndrome. Crit Care
Med. 2018;46(4):619–24.
3. In our main analysis, we used oxygenation index
(OI) calculated on the second day of PARDS, which
Publisher’s Note
has strong correlation to clinical outcomes [3]. Springer Nature remains neutral with regard to jurisdictional claims in
However, anticipating the possibility of PARDS published maps and institutional affiliations.
progression in the following days leading up to
HFOV use, we also performed a sensitivity analysis
using daily OI to adjust for time-varying confound-
ing in the first week of PARDS (Tables S7.1 and
S7.2). All three statistical approaches (GM, propen-
sity score matching and marginal structural model-
ling) demonstrate a consistent direction of harmful
effect. This approach is likely more robust than
using a single day’s OI value.

In summary, our findings remain consistent—there is

a signal toward harm in using HFOV in the general co-
hort of PARDS patients, especially in non-severe cases
who otherwise have no other organ involvement. For the
subgroup with severe PARDS, an empirical study is ne-
cessary and warranted to determine the effects of
HFOV.

Acknowledgements
None

Authors’ contributions
HK and MCJK were both involved in the conception of the article. HK wrote
the first draft. HK and MCJK revised further drafts and read and approved the
final manuscript.