CONTEXT: Inhaled corticosteroids (ICS) are associated with an increased risk of pneumonia in adult patients with chronic obstructive pulmonary disease.
OBJECTIVE: To assess the association between ICS use and risk of pneumonia and other respiratory infections in children with asthma.
DATA SOURCES: We searched PubMed from inception until May 2015. We also searched clinicaltrials.gov and databases of pharmaceutical manufacturers.
STUDY SELECTION: We selected randomized trials that compared ICS with placebo for at least 4 weeks in children with asthma.
DATA EXTRACTION: We included 39 trials, of which 31 trials with 11 615 patients contributed data to meta-analyses.
RESULTS: The incidence of pneumonia was 0.58% (44/7465) in the ICS group and 1.51% (63/4150) in the placebo group. The meta-analysis of 9 trials that revealed at least 1 event of pneumonia revealed a reduced risk of pneumonia in patients taking ICS (risk ratio [RR]: 0.65; 95% confidence interval [CI]: 0.44 to 0.94). Using risk difference as effect measure, the meta-analysis including all 31 trials revealed no significant difference in the risk of pneumonia between the ICS and placebo groups (risk difference: –0.1%; 95% CI: –0.3% to 0.2%). No significant association was found between ICS and risk of pharyngitis (RR: 1.01; 95% CI: 0.87 to 1.18), otitis media (RR: 1.07; 95% CI: 0.83 to 1.37), and sinusitis (RR: 0.89; 95% CI: 0.76 to 1.05).
LIMITATIONS: Lack of clearly defined criteria for respiratory infections and possible publication bias.
CONCLUSIONS: Regular use of ICS may not increase the risk of pneumonia or other respiratory infections in children with asthma.
- CI —
- confidence interval
- ICS —
- inhaled corticosteroids
- ITT —
- OR —
- odds ratio
- RD —
- risk difference
- RR —
- risk ratio
Inhaled corticosteroids (ICS) are widely used in the treatment of asthma and chronic obstructive pulmonary disease.1,2 They are generally considered safe and well tolerated in both adults and children.3,4 However, recent studies have raised concerns about increased risk of pneumonia related to regular use of ICS in adult patients with chronic obstructive pulmonary disease.5–7 There have been few studies assessing the association between ICS and risk of pneumonia in patients with asthma.
The retrospective analysis of individual patient data from 26 AstraZeneca-sponsored randomized trials revealed a reduced risk of pneumonia in patients (4 to 78 years) with asthma treated with budesonide.8 A recent meta-analysis of 10 randomized trials also revealed a protective effect of ICS against pneumonia in patients with asthma.9 However, caution should be taken in interpreting the results of this review. Firstly, “pneumonia” and “pneumonitis” were used as the search terms. Such search strategy might exclude a considerable number of relevant studies given that pneumonia is not a well-established adverse event of ICS and is less likely to be used as keyword or subject heading in clinical trials. Secondly, the authors limited literature search to adults, but the largest trial that contributed with 83.1% of the weight included patients aged 5 to 66 years of age. No separate data were available for pediatric patients.
One recent study revealed an increased risk of oropharyngeal colonization by Streptococcus pneumoniae in children with asthma regularly taking ICS.10 An increased microbial burden in the oropharyngeal area may lead to a higher risk of respiratory infections, especially infections of neighborhood sites such as pharyngitis, otitis media, and sinusitis. Up until now, no study has assessed the possible link between use of ICS and the risk of other respiratory infections besides pneumonia in patients with asthma. Thus, we conducted this systematic review and meta-analysis of randomized trials to assess the association between regular use of ICS and risk of pneumonia, as well as other respiratory infections, in children with asthma. We also investigated the potential effect modifiers, such as drug molecules, daily doses, delivery devices, duration of treatment, and patient’s age.
We followed the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement11 for writing this systematic review. The review protocol was registered on PROSPERO (the International Prospective Register of Systemic Reviews, CRD42015020656).12 There are 2 differences between the previously registered protocol and this review: “triamcinolone” was added in addition as a search term, and 5 posthoc subgroup analyses and 2 posthoc sensitivity analyses were also added.
We searched PubMed from inception until May 2015 to identify clinical trials. The search strategy was as follows: asthma AND (“inhaled corticosteroid” OR “inhaled steroid” OR beclomethasone OR budesonide OR ciclesonide OR flunisolide OR fluticasone OR mometasone OR triamcinolone). There was no restriction on language of publication. We also identified systematic reviews of randomized trials that compared ICS with placebo, by searching PubMed and the Virtual Health Library of the Latin American and Caribbean Center on Health Sciences Information, which contains Medline, Cochrane Central Register of Controlled Trials (CENTRAL), Literatura Latino-Americana e do Caribe em Ciencias da Saude (LILACS), Scientific Electronic Library OnLine (SciELO), the Cochrane Library, and more than 20 other databases.13 Reference lists of primary studies and systematic reviews were screened for additional relevant trials. We also conducted searches on clinicaltrials.gov and the databases of 2 main pharmaceutical manufacturers of ICS (GSK and AstraZeneca) to identify unpublished trials and to obtain unpublished safety data of the trials.
To be included in this review, studies had to meet all of the following criteria: (1) study design: randomized controlled trials; (2) participants: children up to 18 years old with diagnosis of asthma; (3) interventions and comparisons: daily use of ICS, delivered by any type of inhalation device for at least 4 weeks, compared with placebo delivered by the same type of device; (4) outcome measures: at least 1 of respiratory infections reported as adverse event. For this review, we defined “clinically diagnosed pneumonia, with or without radiological confirmation” as the primary outcome. The secondary outcomes included other respiratory infections, such as pharyngitis, otitis media, sinusitis, bronchitis, bronchiolitis, and influenza.
We excluded cross-over trials, trials that compared ICS to other interventions without placebo, trials that used ICS plus other drugs, and trials that included children but in which pediatric patient data were not identifiable.
Four review authors (CC, CS, SM, VM), divided into 2 groups, independently assessed the titles and abstracts of all citations identified by the searches. We obtained the full articles when they appeared to meet the inclusion criteria or there were insufficient data in the title and abstract to make a clear decision for their inclusion. The definitive inclusion of trials was made after reviewing the full-text articles. Any disagreements about study selection were resolved by discussion and, where necessary, a senior reviewer (LZ) was consulted.
Data Extraction and Management
Data from the included trials were extracted independently by 4 review authors (CC, CS, SM, VM) and cross-checked. A standardized data extraction form was used to extract the following data: (1) study characteristics: year of publication, country, and setting of study; (2) methods: study design, methods of random sequence generation, allocation concealment and blinding, description of withdrawal, and adherence to treatment; (3) participants: sample size, age, sex, and inclusion and exclusion criteria; (4) interventions: corticosteroid molecule, daily dose, interval of administration, duration of treatment, drug delivery device, and cointerventions; (5) outcomes: we extracted the number of participants affected and the total number of participants in each treatment arm. For trials that examined multiple doses of ICS, we combined the active treatment groups. Intention-to-treat (ITT) data sets were used whenever available. Given that it is likely that outcomes for which no events occur in either arm may not be mentioned in reports of many randomized trials, we considered no events of pneumonia in both arms if the trial had revealed at least 1 respiratory infection event but not explicitly reported pneumonia. We obtained additional or unpublished safety data through Clinical Trial Register (clinicaltrials.gov) and databases of pharmaceutical manufacturers. When there was inconsistency between unpublished data and data from published articles, we used only unpublished data from above-mentioned databases because they generally provide more detailed and accurate information about drug adverse events.
Assessment of Risk of Bias in Included Studies
Four reviewers (CC, CS, SM, VM) independently assessed the risk of bias in included trials by examining the 6 key domains according to the recommendations of the Cochrane Collaboration14: (1) allocation sequence generation, (2) concealment of allocation, (3) blinding of participants and investigators, (4) incomplete outcome data, (5) selective outcome reporting, and (6) other sources of bias. We graded each potential source of bias as yes, no, or unclear, relating to whether the potential for bias was low, high, or unknown.
Any disagreements about assessment of risk of bias were resolved by discussion and, where necessary, a senior reviewer (LZ) was consulted.
Data Synthesis and Statistical Analysis
We performed meta-analyses for quantitative data synthesis. The risk ratios (RRs) and 95% confidence intervals (CIs) were used to estimate the risk of adverse events associated with ICS. We used the random-effects model for pooling RRs and 95% CIs, assuming that the patients and interventions in clinical trials would have differed in ways that could have impacted on the results, and therefore the trials would not share a common effect size.14 For trials with no events in 1 arm, 0.5 was added to each of the 4 cell counts of 2-by-2 tables. Trials with no events in both arms were not included in the pooled analysis. Given that many outcomes are rare events, we also undertook meta-analyses by using Peto’s odds ratios (ORs) as effect size measure. For the primary outcome pneumonia, we also used risk difference (RD) as effect size measure because RD method appears to have advantage over relative statistics (RR, OR) in that the RD is defined as zero when no events occur in either arm.14 Such trials are therefore included in the pooled analysis.
We assessed heterogeneity in results between trials by using the Cochrane Q test (P < .1 considered significant) and the I2 statistic.15
We conducted posthoc subgroup analyses on the basis of the type of drugs, type of delivery devices (metered-dose inhaler, dry powder inhaler, nebulizer), duration of treatment (≥6 months, <6 months), patient’s age (<4 years, 4 to 18 years), and data source (published data, unpublished data). The χ2 test was used for subgroup comparison. To assess dose–response effects of ICS, we conducted a meta-analysis including only trials that compared directly 2 or more daily doses. We conducted 2 posthoc sensitivity analyses, excluding the trials in which adverse events were not collected by systematic assessment and the trials that contributed more than 50% of the weight to the meta-analysis. All meta-analyses were performed by using Stata version 11.0 (Stata Corp, College Station, TX).
Literature Search and Study Selection
The search strategy identified 2756 unique records from PubMed. After screening the titles and abstracts, we retrieved 227 potentially relevant full-text articles for further evaluation. One hundred fifty-four articles were excluded for reasons shown in Fig 1. We identified 7 unpublished trials from clinicaltrials.gov. Twelve additional trials were found by checking the reference lists of 9 systematic review articles. Thus, a total of 92 randomized trials comparing ICS with placebo in children with asthma were selected. Thirty-seven trials did not reveal safety data, and 16 trials did not reveal respiratory infections as adverse events. Thirty-nine trials16–54 involving 13 595 (8945 ICS-treated, 4650 placebo) children with persistent asthma were therefore included in the review. Thirty-one trials16–46 involving 11 615 participants (7465 ICS-treated, 4150 placebo) contributed data to the meta-analyses.
Study Characteristics and Risk of Bias
Table 1 summarizes the characteristics of the 39 included trials. Thirty-two were published trials,16–39,47–54 but unpublished safety data were obtained through clinicaltrials.gov and databases of pharmaceutical manufacturers in 7 trials.16,20,23,26,27,30,39 Seven were unpublished trials identified through clinicaltrials.gov.40–46 All 39 studies were randomized, double-blind, placebo-controlled trials. Thirty-five16–20,22–35,37–46 were multicenter trials. Only 6 trials16–18,35–37 were designed to assess the safety of ICS, whereas the remaining 33 trials aimed mainly to assess the efficacy of ICS. All but 4 trials24,28,48,50 were industry funded. Five ICS (beclomethasone, budesonide, ciclesonide, fluticasone, and mometasone) given at low or medium daily doses were used. Duration of intervention varied from 4 weeks to 3 years. Adverse events were collected by systematic assessment in 29 trials,16,18,20,22–25,29–39,41–43,45–47,50–54 but none of the trials explicitly defined the criteria for diagnosis of pneumonia and other respiratory infections. All 39 trials were stated as randomized; however, only 9 published trials18,20,28,29,31,33,35,50,51 described the methods for random sequence generation and 228,35 described the methods for allocation concealment (Supplemental Table 4). The risk of performance and detection bias was low in all 39 trials because matching placebos were used for blinding. The risk of attrition bias was also low as ITT or quasi-ITT population was used for calculating rate of adverse events in all 31 trials that contributed data to the meta-analyses. Given that respiratory infections are not well-established adverse events of ICS, underreporting of some respiratory infection outcomes might occur. We rated the risk of outcome reporting bias as high in the 3 trials21,28,35 because only 1 type of respiratory infection was reported.
Use of ICS and Risk of Pneumonia
Nine trials23,27,30,38,40,41,43,44,46 involving a total of 4617 participants (2684 ICS-treated, 1933 placebo) revealed at least 1 event of pneumonia. The meta-analysis of 9 trials revealed that ICS significantly reduced the risk of pneumonia compared with placebo (RR: 0.65, 95% CI: 0.44 to 0.94, P = .02, I2 = 0%; Peto’s OR: 0.63, 95% CI: 0.43 to 0.93, P = .02, I2 = 22.1%; Table 2). Posthoc subgroup analyses did not reveal significant effects of drug molecules, delivery devices, treatment duration, patient’s age, and data source on the size of risk of pneumonia related to ICS (Table 3). The posthoc sensitivity analysis excluding 1 trial38 that contributed more than 50% of the weight revealed no significant association between use of ICS and risk of pneumonia (RR: 0.62, 95% CI: 0.21 to 1.86, P = .39, I2 = 0%; Peto’s OR: 0.57, 95% CI: 0.17 to 1.89, P = .36, I2 = 31.6%). In another posthoc sensitivity analysis, excluding 3 trials27,40,44 in which adverse events were not collected by systematic assessment did not significantly change the results.
Using RD as effect estimate, we conducted the meta-analysis including all 31 trials that revealed at least 1 respiratory infection event. Twenty-three trials16–22,24–26,28,29,31–37,39,42,45,46 did not explicitly reveal pneumonia as adverse events, and we considered no events of pneumonia in both treatment arms. Thus, 44 patients in the corticosteroid group (44/7465, 0.58%) and 63 patients in the placebo group (63/4150, 1.51%) had pneumonia as adverse events. The summary RD between the corticosteroid and placebo groups was –0.1% (95% CI: –0.3% to 0.2%, P = .72, I2 = 0%; Table 2). The posthoc sensitivity analysis excluding 8 trials17,19,21,26–28,40,44 in which adverse events were not collected by systematic assessment yielded the similar results.
Use of ICS and Risk of Other Respiratory Infections
Figure 2 reveals the results of meta-analyses of secondary outcomes—other respiratory infections. No significant association was found between use of ICS and risk of pharyngitis (23 trials,16,18–20,22,23,26,27,29,31,33–42,44–46 RR: 1.01, 95% CI: 0.87 to 1.18, I2 = 19%), otitis media (17 trials,16,18–20,23,26,27,30,32,33,37–41,44,46 RR: 1.07, 95% CI: 0.83 to 1.37, I2 = 34.2%), sinusitis (15 trials,16,18,22,25,26,29,30,34,37–41,44,46 RR: 0.89, 95% CI: 0.76 to 1.05, I2 = 0%), bronchitis (9 trials,16,23,27,28,32,37–39,46 RR: 0.89, 95% CI: 0.75 to 1.05, I2 = 0%), and influenza (4 trials,16,23,41,45 RR: 1.12, 95% CI: 0.65 to 1.93, I2 = 0%). Posthoc subgroup analyses did not reveal significant effects of drug molecules, delivery devices, treatment duration, patient’s age, and data source on the treatment effect size of ICS (Table 3).
Use of ICS and Risk of Respiratory Infections: Dose–Response Relationship
Seventeen trials16,18,19,22,23,29–31,34–37,39,41–43,46 in which 2 or more daily doses of ICS were compared had contributed data to the meta-analyses. Five trials16,23,30,39,46 used fluticasone (100 vs 200 μg/day), 3 trials18,34,43 used budesonide (200 to 500 vs 400 to 1000 μg/day), 1 trial29 used beclomethasone (100 vs 200 μg/day), 3 trials19,36,37 used mometasone (100 vs 200 μg/day), and 5 trials22,31,35,41,42 used ciclesonide (50 to 100 vs 100 to 200 μg/day). Three trials compared 3 daily doses, of which 2 trials22,31 used ciclesonide (50, 100, and 200 μg/day), and 1 trial36 used mometasone (200, 400, and 800 μg/day). In these 3 trials, we combined 2 higher doses into a single group. The meta-analyses did not reveal a significant dose–response relationship between ICS and risk of respiratory infections (Fig 3). Moreover, an inverse relationship was observed between dose of ICS and risk of otitis media, that is, higher dose of ICS was significantly associated with lower risk of otitis media (9 trials, RR: 0.73, 95% CI: 0.57 to 0.93, I2 = 0%).
This systematic review and meta-analysis of randomized trials suggests that regular use of ICS is unlikely to increase the risk of pneumonia or other respiratory infections in children with asthma. Drug molecules, daily doses, delivery devices, treatment duration, and patient’s age appear to have no significant impact on the size of risk of respiratory infections related to use of ICS.
In this review, we used 2 data sets and 3 different effect size measures (RR, Peto’s OR, and RD) to estimate the risk of pneumonia due to ICS. The first data set includes only 9 trials that revealed at least 1 event of pneumonia. The meta-analysis of 9 trials using RR as effect measure reveals that use of ICS could result in a 35% reduction in the risk of pneumonia when compared with placebo. The meta-analysis using Peto’s OR as effect measure yields a similar result. These results are consistent with those found in a meta-analysis9 of 10 trials in which the majority of participants were adults with asthma. However, caution should be taken when interpreting these results. Firstly, meta-analysis may overestimate the incidence of pneumonia if excluding the trials in which no events of pneumonia were reported in either arm. Secondly, 1 trial38 contributed with 89% of the weight to the meta-analysis and the results were no longer statistically significant when this trial was excluded in the sensitivity analysis. However, exclusion of this trial changed the upper limit of 95% CI from 0.94 to 1.86, suggesting a potential increase of risk of pneumonia on the basis of the results of smaller trials. Thirdly, asthma exacerbations could be easily misclassified as pneumonia given the lack of clearly defined criteria for pneumonia in the trials and similar clinical manifestations between pneumonia and asthma exacerbation.
The second data set consists of all 31 trials that revealed at least 1 respiratory infection event, even if pneumonia was not explicitly mentioned. Using RD as effect measure, the meta-analysis of 31 trials reveals no significant association between use of ICS and risk of pneumonia in children with asthma. In this database, we considered no events in both arms if pneumonia had not been explicitly reported. The rationale for this strategy is that no events of pneumonia in either treatment arm were likely not to be mentioned in the trial reports because the majority of trials were designed to assess the efficacy of ICS and pneumonia was rarely considered an a priori safety end point.
On the other hand, if the trial revealed at least 1 respiratory infection event, it is unlikely that pneumonia, a more severe respiratory infection, could be ignored. Taking all together, we believe that the meta-analysis on the basis of second data set may provide more accurate estimates, suggesting no significant association between use of ICS and risk of pneumonia in children with asthma. We recognize that the method used for dealing with possibly no events in both arms is not yet validated, but no other better method is currently available.
ICS may affect local immunity and microbial community in the respiratory tract.55 One cross-sectional study demonstrates that children with asthma regularly taking ICS are almost 4 times more likely to have oropharyngeal colonization by S pneumoniae than those not taking such drugs when adjusted for potential confounders.10 It could be expected that children with an increased microbial burden in the oropharyngeal area may be at higher risk for respiratory infections, not only pneumonia but also other respiratory infections. However, this systematic review does not find a significantly increased risk of other respiratory infections such as pharyngitis, otitis media, sinusitis, bronchitis, and influenza in children with asthma taking ICS. These findings suggest that impaired local immunity and increased oropharyngeal bacterial colonization related to use of ICS may not be associated with an increased risk of respiratory infections in children with asthma.
We also explored other potential effect modifiers through subgroup analyses, and we did not find significant impact of drug molecules, delivery devices, treatment duration, and patient’s age on the effect size of ICS.
The meta-analysis of 17 trials that compared different doses of ICS does not reveal a dose–response relationship between ICS and risk of respiratory infections in children with asthma. However, only low to medium doses of ICS were used in the trials.
Most of the included trials did not describe the methods for random sequence generation and allocation concealment. This may raise a concern about the potential selection bias. However, these potential biases may have no significant impact on the trial results given that the baseline characteristics were similar between treatment groups in all included trials. Underreporting of adverse events of treatment is common in clinical trials. This inherent methodological shortcoming of primary trials is the main limitation of all systematic review of safety data. It may underestimate the rate of adverse events; however, the risk estimation could not be substantially affected because underreporting may occur equally in both treatment arms. To minimize outcome reporting bias, we obtained additional data for 7 published trials through clinicaltrials.gov. Another inherent methodological shortcoming of primary trials is lack of clearly defined criteria for diagnosis of respiratory infections and this should be considered as the other main limitation of this review. Publication bias is always a concern in meta-analysis and there is no standard method for detecting it.56 To minimize the potential publication bias, we have attempted to identify the maximum number as possible of published and unpublished trials through the following strategies: building an effective search strategy by using only 2 types of search terms (asthma and inhaled corticosteroids); checking the references of retrieved articles including systematic reviews to find additional relevant trials; searching clinicaltrials.gov and the databases of GSK and AstraZeneca to identify unpublished trials. Moreover, the subgroup analysis on the basis of data source did not find significant difference in the results between published and unpublished data, suggesting that the potential publication bias might not substantially affect the results of the review. Given that this review did not include trials that used ICS plus other drugs, caution should be taken when extrapolating the findings of this review to patients taking ICS plus long-acting β2 agonist or other drugs. Children aged 12 to 16 through 18 years might be underrepresented because only 6 trials of this review included children in this age group. Some posthoc subgroup analyses and meta-analysis of dose–response might be underpowered to detect an effect. Use of antibiotic and oral steroids may have acted as a potential confounder because patients taking placebo were more likely to receive these medications than those taking ICS due to poorer asthma control. Unfortunately, there was no available data from the trials to assess the effect of potential confounding.
The quality of evidence provided by this systematic review could be graded only as low for primary outcome (pneumonia) and moderate for secondary outcomes (other respiratory infections), according to the Grading of Recommendations, Assessment, Development and Evaluations57 criteria (Supplemental Table 5). Further prospective studies examining the safety of ICS in children with asthma including well-defined pneumonia and other respiratory infections as a priori outcomes are still needed.
The evidence provided by this systematic review suggests that regular use of ICS is unlikely to increase the risk of pneumonia or other respiratory infections in children with asthma. These data add to the already considerable body of evidence suggesting a good safety profile of ICS in children with asthma.
- Accepted December 2, 2016.
- Address correspondence to Linjie Zhang, MD, PhD, Associate Professor, Faculty of Medicine, Federal University of Rio Grande, Rua Visconde de Paranagua 102, Centro, Rio Grande-RS, Brazil. E-mail:
Ms Cazeiro conceptualized and designed the study, participated in trial selection, quality assessment, data collection, data analysis and interpretation, and drafted the protocol and the review article; Dr Silva, Ms Mayer, and Dr Mariany provided input for study conception and design, participated in trial selection, quality assessment and data collection, and critically revised the manuscript; Dr Wainwright provided input for study conception and design and critically revised the manuscript; Dr Zhang contributed to study conception and design, participated in trial selection, quality assessment, data collection, data analysis and interpretation, and critically revised the review protocol and the manuscript; and all authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: Dr Wainwright declared the following relationships/conditions/circumstances that may present a potential conflict of interest: income on a per patient basis derived from pharmaceutical studies (Vertex Pharmaceuticals, Inc; Boehringer Ingelheim; and Ablynx NV); epidemiological research support from GlaxoSmithKline (analysis of bronchoalveolar lavage [BAL] fluid from children with respiratory disorders, 2010–2012); research grant from Novo Nordisk Pharmaceuticals P/L- CF-IDEA Study, 2012–2013; and honorarium and travel reimbursement from Vertex Pharmaceuticals, Inc, Gilead Sciences, Inc, Medscape, and Novartis Pharmaceuticals; the other authors have indicated they have no potential conflicts of interest to disclose.
- Thomson NC,
- Spears M
- Liberati A,
- Altman DG,
- Tetzlaff J, et al
- Zhang L,
- Cazeiro C,
- Silva C,
- Mayer S,
- Mariany V
- Virtual Health Library of the Latin American and Caribbean Center on Health Sciences Information (BIREME)
- Higgins JPT,
- Green S
- Higgins JPT,
- Thompson SG,
- Deeks JJ,
- Altman DG
- Becker AB,
- Kuznetsova O,
- Vermeulen J, et al; Pediatric Montelukast Linear Growth Study Group
- Levy AL,
- Given JT,
- Weinstein SF, et al
- Skoner DP,
- Meltzer EO,
- Milgrom H,
- Stryszak P,
- Teper A,
- Staudinger H
- Wasserman RL,
- Baker JW,
- Kim KT, et al
- ↵FLTA2007 trial. 2005. Available at: https://gsk.sylogent.com/files/1327.pdf. Accessed January 10, 2015
- NCT00163293 trial
- NCT00392288 trial
- NCT00569192 trial
- NCT00920959 trial
- NCT01136382 trial
- FMS30059 trial
- Baker JW,
- Mellon M,
- Wald J,
- Welch M,
- Cruz-Rivera M,
- Walton-Bowen K
- Connett GJ,
- Warde C,
- Wooler E,
- Lenney W
- Jónasson G,
- Carlsen KH,
- Blomqvist P
- Shapiro G,
- Mendelson L,
- Kraemer MJ,
- Cruz-Rivera M,
- Walton-Bowen K,
- Smith JA
- Copyright © 2017 by the American Academy of Pediatrics