Video Abstract

Video Abstract

Close modal
BACKGROUND AND OBJECTIVES:

Meta-analyses of nebulized hypertonic saline (HS) for acute viral bronchiolitis have yielded disparate conclusions. Trial sequential analysis (TSA) is a novel method designed to account for potential sources of error in conventional meta-analysis. We sought to use TSA to determine if the existing literature base is sufficient to draw firm conclusions about the effectiveness of HS in bronchiolitis.

METHODS:

We used the cohort of studies identified in previously published conventional meta-analyses. Included studies were those in which authors compared treatment with HS versus normal saline (or supportive care) in children with bronchiolitis to reduce hospital length of stay (LOS) or hospitalizations. TSA results are used to provide a required information size and monitoring boundaries for statistical significance.

RESULTS:

For the LOS outcome, 17 studies including 1866 patients analyzed in which authors used conventional meta-analysis reveal a statistically significant benefit (mean difference = −0.41 days; 95% confidence interval = −0.07 to −0.75); however, TSA suggests that those conclusions are premature because of failure to reach the adequate information size of 2665 individuals. For the risk of hospitalization outcome, 8 studies including 1728 patients analyzed in which authors used conventional meta-analysis reveal a reduction in the relative risk of hospitalization (relative risk = 0.86; 95% confidence interval = 0.76 to 0.98); however, TSA suggests these conclusions are premature because of failure to reach the adequate information size of 4770. Both LOS and hospitalization results from conventional meta-analysis would be considered potentially false-positives by TSA.

CONCLUSIONS:

TSA reveals that concluding benefit from HS for children with bronchiolitis potentially represents type I error.

What’s Known on This Subject:

Researchers of numerous clinical trials have examined whether nebulized hypertonic saline in acute viral bronchiolitis reduces hospitalization or shortens hospital length of stay, with diminishing benefit observed over time. However, researchers using conventional meta-analyses still report statistically significant benefit.

What This Study Adds:

Trial sequential analysis reveals that conclusions from conventional meta-analysis about hypertonic saline potentially represent type I error and that authors of further studies may fail to identify a clinically meaningful effect.

Hypertonic saline (HS) has been intensely scrutinized as a therapy for acute viral bronchiolitis, and its use varies widely between centers in the United States.1 There is persistent debate around whether the therapy shortens hospital length of stay (LOS) or decreases risk of hospitalization, the most relevant studied outcomes. The existing evidence base includes conflicting results from individual trials, many of which are underpowered: a scenario in which meta-analysis is often considered the ideal method for achieving clarity.2,7 

Several meta-analyses of HS in bronchiolitis have been published in the past few years, with the authors of the 3 most recent offering alternative interpretations of the evidence despite producing similar point estimates of effect size.4,6 Zhang et al4 conclude that HS is an effective and safe intervention on the basis of overall summary measures of effect showing statistically significant reductions in LOS and hospitalization. In contrast, Heikkilä et al6 conclude that HS offers only limited clinical benefit, despite observing similar overall point estimates as those reported by Zhang et al.4 Finally, in a living systematic review initiated by Badgett et al,5 it is concluded that HS is likely ineffective in reducing LOS but that there is potential evidence of benefit for preventing hospitalization. Their differing conclusions arise from several fundamental disagreements, including differing inclusion and exclusion criteria for the study cohort, differing beliefs about the significance of study heterogeneity, and lack of thresholds for clinically meaningful outcomes that are universally agreed on.4,6,8 

Although typically placed at the pinnacle of the evidence pyramid, a growing literature investigating the risk of generating faulty conclusions from conventional meta-analysis is emerging.9,11 In conventional meta-analysis, it is assumed that each study is an independent, random sample of representative populations, and the evolution of the evidence base over time is not taken into account. Yet, we know from the work of Ioannidis and others that most published early trials are underpowered and suggestive of therapeutic benefit, with results over time tending to regress to a mean that often encompasses the null hypothesis.12,13 In addition to being subject to publication bias, conventional meta-analytic techniques may be inadequate to address small trial bias and study heterogeneity arising from diversity in study populations, protocols, or outcomes. It is estimated that ˃25% of initial meta-analyses are inaccurately used to conclude a statistically significant treatment effect (type I error).14,15 Conversely, among meta-analyses failing to reject the null hypothesis, <20% are adequately powered to do so, yet conventional meta-analytic techniques provide no information as to whether the lack of a statistically significant treatment effect represents a true finding or an insufficient number of individuals studied (type II error).9,15 

Trial sequential analysis (TSA) is a novel method for improving the quality of information provided through meta-analysis, specifically by being used to incorporate the element of time into the analysis to establish the sample size required to confirm the point estimate produced by conventional meta-analysis. TSA has recently been applied to a variety of pediatric topics and, in some cases, reveals conclusions contrary to conventional meta-analysis.16,20 Our aim in this study was to apply TSA to HS as therapy for acute viral bronchiolitis. Our specific question was as follows: is the existing literature base sufficient to draw firm conclusions on the impact of HS on hospital LOS and risk of hospitalization?

TSA starts with the same methods as standard meta-analysis to calculate weighted summary measures of effect and z scores (the number of SDs from the mean) to test for statistical significance. Next, by using methods adapted from interim trial monitoring, TSA is used to explicitly introduce time as part of the analysis with studies added to the analysis in chronological order, illustrated with a z curve revealing how the cumulative z score changes as new studies are incorporated into the analysis sequentially. To reduce type II error, TSA is used to calculate a required sample size for statistical significance (information size) assuming that a meta-analysis should analyze a number of subjects greater than or equal to a single, adequately powered trial. Additionally, TSA is used to construct monitoring boundaries, which represent adjusted z score thresholds greater than the traditional value of 1.96 when there is inadequate information size, reducing the risk of incorrectly rejecting the null hypothesis because of random error when only small early studies are included or repeated significance testing is performed. Adjustment is performed by using a α-spending function used to maintain a cumulative risk of type I error proportional to the progress made toward achieving an adequate information size, such that monitoring boundaries converge with the traditional 95% confidence interval (CI) once a sufficient number of individuals have accrued to the analysis. Monitoring boundaries and required information size are determined by an expected effect size, event rate (dichotomous outcomes only), and a heterogeneity-adjustment factor (D2). D2, or the diversity index, represents the ratio of between-trial variance to total variance and is similar to the inconsistency index (I2) statistic used to quantify heterogeneity in conventional meta-analysis but is not downwardly biased when study weights are unequal.21 With Fig 1, we provide several hypothetical trajectories for TSA z curves with an explanation of their significance. Methods for TSA are described in greater detail elsewhere.15 

FIGURE 1

Hypothetical cumulative z curve and monitoring boundaries in TSA. Conventional and TSA monitoring boundaries are 2 sided with an α = 5%. For clarity in presentation, we only present positive z scores in figures. The classification used here adapts that used by Brok et al.16 

FIGURE 1

Hypothetical cumulative z curve and monitoring boundaries in TSA. Conventional and TSA monitoring boundaries are 2 sided with an α = 5%. For clarity in presentation, we only present positive z scores in figures. The classification used here adapts that used by Brok et al.16 

Close modal

Included studies were randomized clinical trials of HS (3%, 5%, or 7%) versus 0.9% saline or standard care alone, with or without coadministration of bronchodilators, for acute viral bronchiolitis reporting on risk of hospitalization or hospital LOS as an outcome. We identified candidate studies for our analysis first on the basis of their inclusion in a previously published conventional meta-analysis. To identify meta-analyses, we searched PubMed using search terms “hypertonic saline” and “bronchiolitis” limited to meta-analyses and systematic reviews through March 13, 2018.

Studies were incorporated into TSA in chronological order on the basis of the month and year of first publication (online publication used where available). All base assumptions of effect size (mean difference [MD] for LOS; relative risk [RR] for risk of hospitalization) were derived from conventional meta-analysis by using DerSimonian-Laird random effects models. For the hospitalization outcome, we assumed a control group admission rate of 30%, equal to the US median.22 Given the trend toward the null in published summary statistics over time, sensitivity analyses were also performed evaluating proposed minimum clinically important effect sizes of −0.25 hospital days and 10% absolute risk reduction in hospitalization. We also specified a 2-sided type I (α) error rate of 5% and a power of 80% (type II [β] error rate of 20%). Data were independently extracted by 2 authors (W.H., S.L.R., and/or S.H.). Analysis was performed by using TSA Viewer version 0.9.5.10 β (Copenhagen Trial Unit, Copenhagen, Denmark).

With our initial search, we identified 30 references. After title, abstract, and full-text review, we identified 7 meta-analyses meeting our inclusion and exclusion criteria. We specifically excluded meta-analyses in which authors did not report on both of our study outcomes, LOS and hospitalization.2,3 We then compared the analytic cohorts for each meta-analysis, cross-referencing all studies included in the 7 meta-analyses by year of publication (or online availability) and the search dates provided in each meta-analysis (Tables 1 and 2). We noted discrepancies between all of the study cohorts making it difficult to identify a “consensus” cohort of studies. Thus, we chose to perform TSA using the cohort of studies assembled by Zhang et al4 because their inclusion and exclusion criteria have been relatively consistently applied across multiple updates over time. The data extracted and characteristics of the studies included in the TSA are detailed in Supplemental Tables 3 and 4.

TABLE 1

Agreement in Study Inclusion Across Conventional Meta-analyses on HS and Hospital LOS in Bronchiolitis

Trial (Date of First Availability of Data, Online or in Print)Meta-analysis (Date Literature Search Was Performed) and Summary Measure of Effect (95% CI)
Zhang et al23,aChen et al7 Zhang et al24 Zhang et al25,bBadgett et al5 Heikkilä et al6 Zhang et al4 
June 2010January 2013May 2013May 2015April 2017June 2017August 2017
MD = −1.16 d (−1.55 to −0.77)MD = −0.96 d (−1.38 to −0.54)MD = −1.15 d (−1.49 to −0.82)MD = −0.45 d (−0.82 to −0.08)MD = −0.41 d (−0.71 to −0.12)MD = −0.47 dc (−0.77 to −0.18)MD = −0.41 d (−0.75 to −0.07)
Mandelberg et al26 (February 2003) 
Tal et al27 (March 2006) 
Kuzik et al28 (September 2007) 
Luo et al29 (August 2009) 
Al-Ansari et al30 (June 2010) — Considered outpatient setting Considered outpatient setting Considered outpatient setting 
Luo et al31 (July 2010) — 
Miraglia Del Giudice et al32 (April 2012) — Not stated 
NCT01238848833 (August 2012) — Unpublished data excluded Unpublished data excluded Unpublished data excluded 
Sharma et al34 (December 2012) — Not stated Not stated 
Mahesh Kumar et al35 (February 2013) — — Not indexed in searched databases Not indexed in searched databases 
Pandit et al36 (June 2013) — — — 
Nenna et al37 (November 2013) — — — Evaluates other drug Evaluates other drug Evaluates other drug 
Ojha et al38 (January 2014) — — — 
Wu et al39 (May 2014) — — — Includes admitted patients from ED study 
Teunissen et al40 (June 2014) — — — 
Tinsa et al41 (November 2014) — — — 
Everard et al42 (December 2014) — — — 
Flores et al43 (September 2015) — — — — 
Silver et al44,d (November 2015) — — — Previous wheezing Previous wheezing 
Köse et al45 (March 2016) — — — — 
Ratajczyk-Pekrul et al46 (August 2016) — — — — Not indexed in searched databases Not indexed in searched databases 
Trial (Date of First Availability of Data, Online or in Print)Meta-analysis (Date Literature Search Was Performed) and Summary Measure of Effect (95% CI)
Zhang et al23,aChen et al7 Zhang et al24 Zhang et al25,bBadgett et al5 Heikkilä et al6 Zhang et al4 
June 2010January 2013May 2013May 2015April 2017June 2017August 2017
MD = −1.16 d (−1.55 to −0.77)MD = −0.96 d (−1.38 to −0.54)MD = −1.15 d (−1.49 to −0.82)MD = −0.45 d (−0.82 to −0.08)MD = −0.41 d (−0.71 to −0.12)MD = −0.47 dc (−0.77 to −0.18)MD = −0.41 d (−0.75 to −0.07)
Mandelberg et al26 (February 2003) 
Tal et al27 (March 2006) 
Kuzik et al28 (September 2007) 
Luo et al29 (August 2009) 
Al-Ansari et al30 (June 2010) — Considered outpatient setting Considered outpatient setting Considered outpatient setting 
Luo et al31 (July 2010) — 
Miraglia Del Giudice et al32 (April 2012) — Not stated 
NCT01238848833 (August 2012) — Unpublished data excluded Unpublished data excluded Unpublished data excluded 
Sharma et al34 (December 2012) — Not stated Not stated 
Mahesh Kumar et al35 (February 2013) — — Not indexed in searched databases Not indexed in searched databases 
Pandit et al36 (June 2013) — — — 
Nenna et al37 (November 2013) — — — Evaluates other drug Evaluates other drug Evaluates other drug 
Ojha et al38 (January 2014) — — — 
Wu et al39 (May 2014) — — — Includes admitted patients from ED study 
Teunissen et al40 (June 2014) — — — 
Tinsa et al41 (November 2014) — — — 
Everard et al42 (December 2014) — — — 
Flores et al43 (September 2015) — — — — 
Silver et al44,d (November 2015) — — — Previous wheezing Previous wheezing 
Köse et al45 (March 2016) — — — — 
Ratajczyk-Pekrul et al46 (August 2016) — — — — Not indexed in searched databases Not indexed in searched databases 
a

Citation was previously indexed as the 2008 version; however, only the updated 2011 version is currently available through Cochrane Review.

b

Internal update is by the same author group as previous Cochrane Reviews but is published in Pediatrics.

c

MD from abstract differs from text.

d

Also identified by trial number NCT01488448.

TABLE 2

Agreement in Study Inclusion Across Conventional Meta-analyses on HS and Hospital Admission Rates in Bronchiolitis

Trial (Date of First Availability of Data, Online or in Print)Meta-analysis (Date Literature Search Was Performed) and Summary Measure of Effect (95% CI)
Zhang et al23,aChen et al7 Zhang et al24 Zhang et al25 Badgett et al5 Heikkilä et al6 Zhang et al4 
June 2010January 2013May 2013May 2015April 2017June 2017August 2017
RR = 0.63 (0.34 to 1.17)RR = 0.59 (0.37 to 0.93)RR = 0.63 (0.37 to 1.07)RR = 0.80 (0.67 to 0.96)RR = 0.79 (0.67 to 0.95)RR = 0.77 (0.62 to 0.96)RR = 0.86 (0.76 to 0.98)
Sarrell et al47 (December 2002) Not stated 
Grewal et al48 (November 2009) 
Anil et al49 (January 2010) 
Kuzik et al50 (November 2010) — Previous wheeze Previous wheeze Not stated Previous wheeze Previous wheeze 
Ipek et al51 (December 2011) — 
Jacobs et al52 (December 2013) — — — 
Florin et al53 (May 2014) — — — 
Wu et al39 (May 2014) — — — 
NCT0204523854 (January 2015) — — — Not stated Unpublished data excluded Awaiting classification 
Khanal et al55 (September 2015) — — — — Did not report outcomeb 
Angoulvant et al56 (June 2017) — — — — — 
Trial (Date of First Availability of Data, Online or in Print)Meta-analysis (Date Literature Search Was Performed) and Summary Measure of Effect (95% CI)
Zhang et al23,aChen et al7 Zhang et al24 Zhang et al25 Badgett et al5 Heikkilä et al6 Zhang et al4 
June 2010January 2013May 2013May 2015April 2017June 2017August 2017
RR = 0.63 (0.34 to 1.17)RR = 0.59 (0.37 to 0.93)RR = 0.63 (0.37 to 1.07)RR = 0.80 (0.67 to 0.96)RR = 0.79 (0.67 to 0.95)RR = 0.77 (0.62 to 0.96)RR = 0.86 (0.76 to 0.98)
Sarrell et al47 (December 2002) Not stated 
Grewal et al48 (November 2009) 
Anil et al49 (January 2010) 
Kuzik et al50 (November 2010) — Previous wheeze Previous wheeze Not stated Previous wheeze Previous wheeze 
Ipek et al51 (December 2011) — 
Jacobs et al52 (December 2013) — — — 
Florin et al53 (May 2014) — — — 
Wu et al39 (May 2014) — — — 
NCT0204523854 (January 2015) — — — Not stated Unpublished data excluded Awaiting classification 
Khanal et al55 (September 2015) — — — — Did not report outcomeb 
Angoulvant et al56 (June 2017) — — — — — 
a

Citation was previously indexed as the 2008 version; however, only the updated 2011 version is currently available through Cochrane Review.

b

Khanal et al report on readiness for discharge rather than traditional admission rates.

Conventional meta-analysis involving 17 studies (including 1 unpublished study)26,29,31,36,38,40,43,45,46 and 1866 patients reveals that HS may reduce hospital LOS (MD = −0.41 days; 95% CI = −0.07 to −0.75) with substantial heterogeneity (I2 = 79%; D2 = 84%). However, under TSA, the z curve fails to cross the monitoring boundary indicating that conclusions based on this point estimate remain at risk for being false-positives and that an additional 799 individuals would need to be studied to achieve the required information size to reach firm conclusions (Fig 2). If the current trend in studies were to continue, and we assumed a minimum clinically important effect size of −0.25 days (6 hours), an information size of 7160 individuals would be required to reach definitive conclusions.

FIGURE 2

Cumulative z curve and monitoring boundaries for LOS outcome. Conventional and TSA monitoring boundaries are 2 sided with an α = 5%. For clarity in presentation, we only present positive z scores in figures. Studies are listed in the order they entered the analysis and are as follows: Mandelberg et al,26 Tal et al,27 Kuzik et al,28 Luo et al,29 Sharma et al,34 Maheshkumar et al,35 Pandit et al,36 Ojha et al,38 and Teunissen et al.40 

FIGURE 2

Cumulative z curve and monitoring boundaries for LOS outcome. Conventional and TSA monitoring boundaries are 2 sided with an α = 5%. For clarity in presentation, we only present positive z scores in figures. Studies are listed in the order they entered the analysis and are as follows: Mandelberg et al,26 Tal et al,27 Kuzik et al,28 Luo et al,29 Sharma et al,34 Maheshkumar et al,35 Pandit et al,36 Ojha et al,38 and Teunissen et al.40 

Close modal

Conventional meta-analysis of 8 studies39,47,49,51,53,56 and 1728 patients reveals that HS may reduce the risk of hospitalization (RR = 0.86; 95% CI = 0.76 to 0.98) with minimal heterogeneity (I2 = 7%; D2 = 25%). However, under TSA, the z curve fails to cross the monitoring boundary indicating that conclusions based on this point estimate remain at high risk of being false-positives and that an additional 3042 patients would need to be studied to achieve the required information size to reach firm conclusions (Fig 3). If the current trend in studies were to continue, and we assumed a minimum clinically important effect size of 10% absolute RR in hospitalization rates (RR = 0.9), an information size of 9469 individuals would be required to reach definitive conclusions.

FIGURE 3

Cumulative z curve and monitoring boundaries for hospitalization outcome. Conventional and TSA monitoring boundaries are 2 sided with an α = 5%. For clarity in presentation, we only present positive z scores in figures. Studies are listed in the order they entered the analysis and are as follows: Sarrell et al,47 Grewal et al,48 Anil et al,49 Ipek et al,51 Jacobs et al,52 Florin et al,53 Wu et al,39 and Angoulvant et al.56 

FIGURE 3

Cumulative z curve and monitoring boundaries for hospitalization outcome. Conventional and TSA monitoring boundaries are 2 sided with an α = 5%. For clarity in presentation, we only present positive z scores in figures. Studies are listed in the order they entered the analysis and are as follows: Sarrell et al,47 Grewal et al,48 Anil et al,49 Ipek et al,51 Jacobs et al,52 Florin et al,53 Wu et al,39 and Angoulvant et al.56 

Close modal

TSA reveals that the required information size to draw firm conclusions from meta-analysis regarding the use of HS in acute viral bronchiolitis has not been achieved to date for either LOS or hospitalization. For these outcomes, ∼70% and 36% of the required sample size to do so have been achieved, respectively. Furthermore, preliminary testing results for statistical significance under TSA are negative despite significance under conventional meta-analysis. Thus, the summary statistics generated from existing meta-analyses are more appropriately treated as hypotheticals rather than definitive conclusions. In this sense, the results obtained under conventional meta-analysis, although not mathematically incorrect, are potentially misleading.

With our results, we suggest that continued skepticism about the effectiveness of HS in children with bronchiolitis is reasonable, and we add further support to the concern that the statistical methods used in most meta-analyses can lead to inflated type I error rates.9,12 This concern is supported by the pattern revealed in this cohort of studies, in which early meta-analyses revealed large treatment effects that diminished as later and larger studies were incorporated. Systematic error can drive false-positive results in meta-analysis because of publication bias, the inclusion of biased or diverse trials, and post hoc selection bias when conducting the analysis.16 TSA is particularly suited for addressing some of these sources of error by being used to identify a required information size representing “sufficient” as opposed to the “best available” evidence, directly account for study heterogeneity, and to explicitly consider evolution of the evidence base over time in the primary analysis.15 

For clinicians seeking to reconcile daily practice decisions with an uncertain evidence base, routine use of HS may be viewed as a trade-off between making an error of omission versus one of commission. The more conservative significance threshold of TSA favors the former over the latter and could delay the use of potentially beneficial therapies. However, given that effect sizes have trended toward the null over time, skepticism is likely warranted in this case. For clinical investigators, an inherent strength of TSA is that it quantifies the necessary size of future studies, a question that conventional meta-analysis cannot be used to answer. Both Zhang et al4 and Heikkilä et al6 state that further clinical trials of nebulized HS are needed. However, given that exceptionally large sample sizes would be necessary to confirm both the currently published effect sizes and those proposed to be of minimum clinical importance, the question of whether more studies truly are needed remains open for debate.

Our study is subject to a number of limitations. First, the findings from any meta-analysis are subject to the strengths, weaknesses, and particular circumstances of the individual trials composing it, an issue our tables outlining the existing study cohorts illustrate clearly by revealing wide variation in individual trial inclusion and exclusion between meta-analyses. Such variability, however, lends further support for employing the more conservative methods for determining statistical significance under TSA. Additionally, we did not perform an independent search for potential studies, instead relying on those identified by previous meta-analyses because part of our aim was to directly compare results from conventional meta-analysis with those obtained by TSA. It is possible that the identification and inclusion of a different group of studies would have affected our results, particularly in regard to whether a required information size had been achieved. Elsewhere, we have previously criticized the choice to interpret meta-analysis at face value when heterogeneity signals are high.2 There was substantial statistical heterogeneity in the LOS cohort that we did not attempt to resolve through sensitivity or subgroup analysis. However, a relative strength of TSA is the explicit incorporation of study heterogeneity adjustments into the analysis. Finally, TSA results depend on making correct assumptions about effect size and event rate. It is important to note that we accepted the empirical estimates of effect size generated by conventional meta-analytic methods as our base assumption; thus, any critique of these assumptions may also be seen as a critique of conventional meta-analysis. Assuming larger effect sizes than the current point estimates would generate smaller required information sizes and likely reveal HS to be effective; however, such assumptions are counterfactual to the existing evidence.

The information size necessary to draw definitive conclusions about the impact of HS on hospital LOS and admission rates in acute viral bronchiolitis has not been achieved to date. Estimates from conventional meta-analysis suggesting clinical benefit may represent false-positives; we also suggest that if the current trend in study results continues, authors of further studies are unlikely to confirm a clinically meaningful effect.

     
  • CI

    confidence interval

  •  
  • D2

    diversity index

  •  
  • HS

    hypertonic saline

  •  
  • I2

    inconsistency index

  •  
  • LOS

    length of stay

  •  
  • MD

    mean difference

  •  
  • RR

    relative risk

  •  
  • TSA

    trial sequential analysis

Dr Harrison performed a portion of the analysis and drafted the manuscript; Dr Angoulvant conceptualized the project and critically reviewed all analyses and the manuscript; Drs House and Gajdos critically reviewed and revised the analyses and the manuscript; Dr Ralston designed the project, performed a portion of the analysis, and drafted the manuscript; and all authors approved the final manuscript as submitted.

FUNDING: No external funding.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2018-1868.

1
Davis
J
,
Thompson
AD
,
Mansbach
JM
, et al
.
Multicenter observational study of the use of nebulized hypertonic saline to treat children hospitalized for bronchiolitis from 2008 to 2014.
Hosp Pediatr
.
2017
;
7
(
8
):
483
491
[PubMed]
2
Maguire
C
,
Cantrill
H
,
Hind
D
,
Bradburn
M
,
Everard
ML
.
Hypertonic saline (HS) for acute bronchiolitis: systematic review and meta-analysis.
BMC Pulm Med
.
2015
;
15
:
148
[PubMed]
3
Brooks
CG
,
Harrison
WN
,
Ralston
SL
.
Association between hypertonic saline and hospital length of stay in acute viral bronchiolitis: a reanalysis of 2 meta-analyses.
JAMA Pediatr
.
2016
;
170
(
6
):
577
584
[PubMed]
4
Zhang
L
,
Mendoza-Sassi
RA
,
Wainwright
C
,
Klassen
TP
.
Nebulised hypertonic saline solution for acute bronchiolitis in infants.
Cochrane Database Syst Rev
.
2017
;
12
:
CD006458
[PubMed]
5
Badgett
RG
,
Vindhyal
M
,
Stirnaman
JT
,
Gibson
CM
,
Halaby
R
.
A living systematic review of nebulized hypertonic saline for acute bronchiolitis in infants.
JAMA Pediatr
.
2015
;
169
(
8
):
788
789
6
Heikkilä
P
,
Renko
M
,
Korppi
M
.
Hypertonic saline inhalations in bronchiolitis-A cumulative meta-analysis.
Pediatr Pulmonol
.
2018
;
53
(
2
):
233
242
[PubMed]
7
Chen
YJ
,
Lee
WL
,
Wang
CM
,
Chou
HH
.
Nebulized hypertonic saline treatment reduces both rate and duration of hospitalization for acute bronchiolitis in infants: an updated meta-analysis.
Pediatr Neonatol
.
2014
;
55
(
6
):
431
438
[PubMed]
8
Haidich
AB
.
Meta-analysis in medical research.
Hippokratia
.
2010
;
14
(
suppl 1
):
29
37
[PubMed]
9
Brok
J
,
Thorlund
K
,
Gluud
C
,
Wetterslev
J
.
Trial sequential analysis reveals insufficient information size and potentially false positive results in many meta-analyses.
J Clin Epidemiol
.
2008
;
61
(
8
):
763
769
[PubMed]
10
Imberger
G
,
Thorlund
K
,
Gluud
C
,
Wetterslev
J
.
False-positive findings in Cochrane meta-analyses with and without application of trial sequential analysis: an empirical review.
BMJ Open
.
2016
;
6
(
8
):
e011890
[PubMed]
11
Garcia-Alamino
JM
,
Bankhead
C
,
Heneghan
C
,
Pidduck
N
,
Perera
R
.
Impact of heterogeneity and effect size on the estimation of the optimal information size: analysis of recently published meta-analyses.
BMJ Open
.
2017
;
7
(
11
):
e015888
[PubMed]
12
Ioannidis
JPA
.
Why most published research findings are false.
PLoS Med
.
2005
;
2
(
8
):
e124
[PubMed]
13
Alahdab
F
,
Farah
W
,
Almasri
J
, et al
.
Treatment effect in earlier trials of patients with chronic medical conditions: a meta-epidemiologic study.
Mayo Clin Proc
.
2018
;
93
(
3
):
278
283
[PubMed]
14
Thorlund
K
,
Devereaux
PJ
,
Wetterslev
J
, et al
.
Can trial sequential monitoring boundaries reduce spurious inferences from meta-analyses?
Int J Epidemiol
.
2009
;
38
(
1
):
276
286
[PubMed]
15
Thorlund
K
,
Engstrom
J
,
Wetterslev
J
, et al
. Copenhagen Trials Unit. Trial sequential analysis.
2017
. Available at: www.ctu.dk/tools-and-links/trial-sequential-analysis.aspx. Accessed March 30, 2018
16
Brok
J
,
Thorlund
K
,
Wetterslev
J
,
Gluud
C
.
Apparently conclusive meta-analyses may be inconclusive–trial sequential analysis adjustment of random error risk due to repetitive testing of accumulating data in apparently conclusive neonatal meta-analyses.
Int J Epidemiol
.
2009
;
38
(
1
):
287
298
[PubMed]
17
Kelley
GA
,
Kelley
KS
,
Pate
RR
.
Exercise and BMI in overweight and obese children and adolescents: a systematic review and trial sequential meta-analysis.
Biomed Res Int
.
2015
;
2015
:
704539
18
Storebø
OJ
,
Krogh
HB
,
Ramstad
E
, et al
.
Methylphenidate for attention-deficit/hyperactivity disorder in children and adolescents: Cochrane systematic review with meta-analyses and trial sequential analyses of randomised clinical trials.
BMJ
.
2015
;
351
:
h5203
[PubMed]
19
Zhang
ZQ
,
Zhong
Y
,
Huang
XM
,
Du
LZ
.
Airway administration of corticosteroids for prevention of bronchopulmonary dysplasia in premature infants: a meta-analysis with trial sequential analysis.
BMC Pulm Med
.
2017
;
17
(
1
):
207
[PubMed]
20
Holmskov
M
,
Storebø
OJ
,
Moreira-Maia
CR
, et al
.
Gastrointestinal adverse events during methylphenidate treatment of children and adolescents with attention deficit hyperactivity disorder: a systematic review with meta-analysis and trial sequential analysis of randomised clinical trials.
PLoS One
.
2017
;
12
(
6
):
e0178187
[PubMed]
21
Wetterslev
J
,
Thorlund
K
,
Brok
J
,
Gluud
C
.
Estimating required information size by quantifying diversity in random-effects model meta-analyses.
BMC Med Res Methodol
.
2009
;
9
:
86
[PubMed]
22
Bourgeois
FT
,
Monuteaux
MC
,
Stack
AM
,
Neuman
MI
.
Variation in emergency department admission rates in US children’s hospitals.
Pediatrics
.
2014
;
134
(
3
):
539
545
[PubMed]
23
Zhang
L
,
Mendoza-Sassi
RA
,
Wainwright
C
,
Klassen
TP
.
Nebulized hypertonic saline solution for acute bronchiolitis in infants.
Cochrane Database Syst Rev.
2008
;(
4
):
CD006458
24
Zhang
L
,
Mendoza-Sassi
RA
,
Wainwright
C
,
Klassen
TP
.
Nebulised hypertonic saline solution for acute bronchiolitis in infants.
Cochrane Database Syst Rev
.
2013
;(
7
):
CD006458
25
Zhang
L
,
Mendoza-Sassi
RA
,
Klassen
TP
,
Wainwright
C
.
Nebulized hypertonic saline for acute bronchiolitis: a systematic review
.
Pediatrics
.
2015
;
136
(
4
):
687
701
26
Mandelberg
A
,
Tal
G
,
Witzling
M
, et al
.
Nebulized 3% hypertonic saline solution treatment in hospitalized infants with viral bronchiolitis.
Chest
.
2003
;
123
(
2
):
481
487
[PubMed]
27
Tal
G
,
Cesar
K
,
Oron
A
,
Houri
S
,
Ballin
A
,
Mandelberg
A
.
Hypertonic saline/epinephrine treatment in hospitalized infants with viral bronchiolitis reduces hospitalization stay: 2 years experience.
Isr Med Assoc J
.
2006
;
8
(
3
):
169
173
[PubMed]
28
Kuzik
BA
,
Al-Qadhi
SA
,
Kent
S
, et al
.
Nebulized hypertonic saline in the treatment of viral bronchiolitis in infants.
J Pediatr
.
2007
;
151
(
3
):
266
270, 270.e1
29
Luo
Z
,
Liu
E
,
Luo
J
, et al
.
Nebulized hypertonic saline/salbutamol solution treatment in hospitalized children with mild to moderate bronchiolitis.
Pediatr Int
.
2010
;
52
(
2
):
199
202
30
Al-Ansari
K
,
Sakran
M
,
Davidson
BL
,
El Sayyed
R
,
Mahjoub
H
,
Ibrahim
K
.
Nebulized 5% or 3% hypertonic or 0.9% saline for treating acute bronchiolitis in infants
.
J Pedatr
.
2010
;
157
(
4
):
630
634
[PubMed]
31
Luo
Z
,
Fu
Z
,
Liu
E
, et al
.
Nebulized hypertonic saline treatment in hospitalized children with moderate to severe viral bronchiolitis.
Clin Microbiol Infect
.
2011
;
17
(
12
):
1829
1833
32
Miraglia Del Giudice
M
,
Saitta
F
,
Leonardi
S
, et al
.
Effectiveness of nebulized hypertonic saline and epinephrine in hospitalized infants with bronchiolitis.
Int J Immunopathol Pharmacol
.
2012
;
25
(
2
):
485
491
[PubMed]
33
NCT01238848. Efficacy of nebulized hypertonic saline in the treatment of acute bronchiolitis (hypertonic). Available at: https://clinicaltrials.gov/ct2/show/NCT01238848. Accessed March 2, 2018
34
Sharma
BS
,
Gupta
MK
,
Rafik
SP
.
Hypertonic (3%) saline vs 0.93% saline nebulization for acute viral bronchiolitis: a randomized controlled trial.
Indian Pediatr
.
2013
;
50
(
8
):
743
747
[PubMed]
35
Maheshkumar
KB
,
Karunakara
BP
,
Nagalli
MM
,
Mallikarjuna
HB
.
Aerosolised hypertonic saline in hospitalized young children with acute bronchiolitis: a randomized controlled clinical trial.
J Pediatr Sci
.
2013
;
5
(
1
):
e174
36
Pandit
S
,
Dhawan
N
,
Thakur
D
.
Utility of hypertonic saline in the management of acute bronchiolitis in infants: a randomised controlled study.
Int J Clin Pediatr
.
2013
;
2
(
1
):
24
29
37
Nenna
R
,
Papoff
P
,
Moretti
C
, et al
.
Seen percent hypertonic saline--0.1% hyaluronic acid in infants with mild-to-moderate bronchiolitis
.
Pediatric Pulmonology
.
2014
;
49
(
9
):
919
925
[PubMed]
38
Ojha
AR
,
Mathema
S
,
Sah
S
,
Aryal
UR
.
A comparative study on use of 3% saline versus 0.9% saline nebulization in children with bronchiolitis.
J Nepal Health Res Counc
.
2014
;
12
(
26
):
39
43
[PubMed]
39
Wu
S
,
Baker
C
,
Lang
ME
, et al
.
Nebulized hypertonic saline for bronchiolitis: a randomized clinical trial.
JAMA Pediatr
.
2014
;
168
(
7
):
657
663
[PubMed]
40
Teunissen
J
,
Hochs
AH
,
Vaessen-Verberne
A
, et al
.
The effect of 3% and 6% hypertonic saline in viral bronchiolitis: a randomised controlled trial.
Eur Respir J
.
2014
;
44
(
4
):
913
921
[PubMed]
41
Tinsa
F
,
Abdelkafi
S
,
Bel Haj
I
, et al
.
A randomized, controlled trial of nebulized 5% hypertonic saline and mixed 5% hypertonic saline with epinephrine in bronchiolitis.
Tunis Med
.
2014
;
92
(
11
):
674
677
[PubMed]
42
Everard
ML
,
Hind
D
,
Ugonna
K
, et al;
SABRE Study Team
.
SABRE: a multicentre randomised control trial of nebulised hypertonic saline in infants hospitalised with acute bronchiolitis.
Thorax
.
2014
;
69
(
12
):
1105
1112
[PubMed]
43
Flores
P
,
Mendes
AL
,
Neto
AS
.
A randomized trial of nebulized 3% hypertonic saline with salbutamol in the treatment of acute bronchiolitis in hospitalized infants.
Pediatr Pulmonol
.
2016
;
51
(
4
):
418
425
[PubMed]
44
Silver
AH
,
Esteban-Cruciani
N
,
Azzarone
G
, et al
.
3% Hypertonic Saline Versus Normal Saline in Inpatient Bronchiolitis: A Randomized Controlled Trial
.
Pediatrics
.
2015
;
136
(
6
):
1036
1043
[PubMed]
45
Köse
S
,
Şehriyaroğlu
A
,
Esen
F
, et al
.
Comparing the efficacy of 7%, 3% and 0.9% saline in moderate to severe bronchiolitis in infants.
Balkan Med J
.
2016
;
33
(
2
):
193
197
[PubMed]
46
Ratajczyk-Pekrul
K
,
Gonerko
P
,
Peregud-Pogorzelski
J
.
The clinical use of hypertonic saline/salbutamol in treatment of bronchiolitis.
Pediatr Pol
.
2016
;
91
(
4
):
301
307
47
Sarrell
EM
,
Tal
G
,
Witzling
M
, et al
.
Nebulized 3% hypertonic saline solution treatment in ambulatory children with viral bronchiolitis decreases symptoms.
Chest
.
2002
;
122
(
6
):
2015
2020
[PubMed]
48
Grewal
S
,
Ali
S
,
McConnell
DW
,
Vandermeer
B
,
Klassen
TP
.
A randomized trial of nebulized 3% hypertonic saline with epinephrine in the treatment of acute bronchiolitis in the emergency department.
Arch Pediatr Adolesc Med
.
2009
;
163
(
11
):
1007
1012
[PubMed]
49
Anil
AB
,
Anil
M
,
Saglam
AB
,
Cetin
N
,
Bal
A
,
Aksu
N
.
High volume normal saline alone is as effective as nebulized salbutamol-normal saline, epinephrine-normal saline, and 3% saline in mild bronchiolitis.
Pediatr Pulmonol
.
2010
;
45
(
1
):
41
47
[PubMed]
50
Kuzik
BA
,
Flavin
MP
,
Zielinski
D
, et al
.
Effect of inhaled hypertonic saline on hospital admission rate in children with viral bronchiolitis: a randomized trial
.
CJEM
.
2010
;
12
(
6
):
477
484
[PubMed]
51
Ipek
IO
,
Yalcin
EU
,
Sezer
RG
,
Bozaykut
A
.
The efficacy of nebulized salbutamol, hypertonic saline and salbutamol/hypertonic saline combination in moderate bronchiolitis.
Pulm Pharmacol Ther
.
2011
;
24
(
6
):
633
637
[PubMed]
52
Jacobs
JD
,
Foster
M
,
Wan
J
,
Pershad
J
.
7% Hypertonic saline in acute bronchiolitis: a randomized controlled trial.
Pediatrics
.
2014
;
133
(
1
). Available at: www.pediatrics.org/cgi/content/full/133/1/e8
[PubMed]
53
Florin
TA
,
Shaw
KN
,
Kittick
M
,
Yakscoe
S
,
Zorc
JJ
.
Nebulized hypertonic saline for bronchiolitis in the emergency department: a randomized clinical trial.
JAMA Pediatr
.
2014
;
168
(
7
):
664
670
[PubMed]
54
NCT02045238.
Inhaled hypertonic saline use in the emergency department to treat acute viral bronchiolitis. Available at: https://clinicaltrials.gov/ct2/show/NCT02045238. Accessed March 2, 2018
55
Khanal
A
,
Sharma
A
,
Basnet
S
,
Sharma
PR
,
Gami
FC
.
Nebulised hypertonic saline (3%) among children with mild to moderately severe bronchiolitis--a double blind randomized controlled trial
.
BMC Pediatr
.
2015
;
15
(
115
):
[PubMed]
56
Angoulvant
F
,
Bellêttre
X
,
Milcent
K
, et al;
Efficacy of 3% Hypertonic Saline in Acute Viral Bronchiolitis (GUERANDE) Study Group
.
Effect of nebulized hypertonic saline treatment in emergency departments on the hospitalization rate for acute bronchiolitis: a randomized clinical trial.
JAMA Pediatr
.
2017
;
171
(
8
):
e171333
[PubMed]

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data