Nebulization of antiinfective agents is a common but unstandardized practice in critically ill patients.
A systematic review of 1,435 studies was performed in adults receiving invasive mechanical ventilation. Two different administration strategies (adjunctive and substitute) were considered clinically relevant. Inclusion was restricted to studies using jet, ultrasonic, and vibrating-mesh nebulizers. Studies involving children, colonized-but-not-infected adults, and cystic fibrosis patients were excluded.
Five of the 11 studies included had a small sample size (fewer than 50 patients), and only 6 were randomized. Diversity of case-mix, dosage, and devices are sources of bias. Only a few patients had severe hypoxemia. Aminoglycosides and colistin were the most common antibiotics, being safe regarding nephrotoxicity and neurotoxicity, but increased respiratory complications in 9% (95% CI, 0.01 to 0.18; I2 = 52%), particularly when administered to hypoxemic patients. For tracheobronchitis, a significant decrease in emergence of resistance was evidenced (risk ratio, 0.18; 95% CI, 0.05 to 0.64; I2 = 0%). Similar findings were observed in pneumonia by susceptible pathogens, without improvement in mortality or ventilation duration. In pneumonia caused by resistant pathogens, higher clinical resolution (odds ratio, 1.96; 95% CI, 1.30 to 2.96; I2 = 0%) was evidenced. These findings were not consistently evidenced in the assessment of efficacy against pneumonia caused by susceptible pathogens.
Performance of randomized trials evaluating the impact of nebulized antibiotics with more homogeneous populations, standardized drug delivery, predetermined clinical efficacy, and safety outcomes is urgently required. Infections by resistant pathogens might potentially have higher benefit from nebulized antiinfective agents. Nebulization, without concomitant systemic administration of the drug, may reduce nephrotoxicity but may also be associated with higher risk of respiratory complications.
In critically ill patients, nebulized antibiotics are increasingly used; however, the safety and efficacy of these are unknown
A systematic review reports that the data are sparse; however, nebulization may be more effective in cases of resistant organisms and less nephrotoxic (if replacing nephrotoxic systemic agents) but may compromise mechanical ventilation especially in hypoxemic patients
AEROSOLIZED administration of antibiotics to treat respiratory infections in critically ill patients was described more than 40 yr ago.1 Supported by experimental studies,2,3 several clinical studies demonstrated that the endotracheal administration of polymyxin B or gentamicin prevented ventilator-associated pneumonia (VAP).4–7 However, when it was prophylactically administered beyond 1 week, the incidence of VAP caused by polymyxin B- resistant pathogens increased,8 leading the critical care community to abandon this administration method. During the 1990s, an enhanced understanding of conditions required for reaching the deep lung during mechanical ventilation,9 together with the development of new-generation nebulizers,10–13 contributed to its reemergence. In 1998, a study performed in tracheostomized patients suggested that the nebulization of aminoglycosides using a jet nebulizer, producing appropriate mass median aerodynamic diameter aerosol particles, was appropriate for treating ventilator-associated tracheobronchitis (VAT).14 Between 2000 and 2010, experimental studies on pharmacokinetics/pharmacodynamics using ultrasonic or vibrating-mesh nebulizers in mechanically ventilated piglets15–18 increased the understanding of several factors influencing nebulization performance, which renewed the interest for antibiotic nebulization.19–23 Respiratory infections have also become more difficult to treat due to greater levels of host immunosuppression and an increasing prevalence of drug-resistant pathogens.24,25 These factors make daily clinical practice highly challenging, with intensivists considering alternative treatment strategies, including nebulization. However, aerosol administration might lead to a potential increased risk of severe adverse events, such as cardiorespiratory or nephrotoxicity.23,26 Given this background, our main aim was to determine the efficacy and safety of this widely extended but yet unstandardized practice by performing a systematic review and meta- analysis of the existing literature. Recent reports describe the complexity and variability of administration practices,27 evidencing the differences in management for VAT and VAP. Our hypothesis was that nebulized antibiotics are safe and effective for therapy of nosocomial respiratory infections in invasively mechanical ventilation adults.
Materials and Methods
This report describes the results of the systematic review and meta-analyses under the guidance of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement.28 A list of clinical questions under the PICO framework (Population-Intervention-Comparison-Outcome) format was created (table 1). The selected population was adult critically ill patients, receiving support with invasive mechanical ventilation, defined as ventilatory support through a nasotracheal tube, orotracheal tube, or tracheostomy. Noninvasive mechanical ventilation or other respiratory support devices such as high-flow nasal therapy were not considered in our study. The respiratory infections considered were VAT, VAP, or severe hospital-acquired pneumonia. Patients who were colonized, with colonization defined as the presence of purulent tracheal secretions without infectious signs and radiologic infiltrate, were excluded from our analyses. The susceptibility pattern of the pathogens was simplified as susceptible or resistant, the latter including bacteria with any type of resistance criteria (multidrug-, extensively drug-, or pandrug-resistant bacteria) defined by the Centers for Disease Control and Prevention (Atlanta, Georgia).29 Nebulization of the antibiotic had to be performed with any device generating particles sufficiently small to reach the lung parenchyma (jet, ultrasonic, or vibrating-mesh nebulizers).
Two different strategies for the administration of nebulized antibiotics were considered clinically relevant for the treatment of VAP:
Adjunctive strategy: nebulized colistin or aminoglycosides administered to patients already receiving IV colistin or aminoglycosides, added to standard first-line IV antibiotics (in comparison to patients also receiving the same IV therapy, but no nebulized antibiotics).
Substitution strategy: nebulized colistin or aminoglycosides administered to patients not receiving IV colistin or aminoglycosides, but only standard first-line IV antibiotics (in comparison to patients receiving IV colistin or aminoglycosides—not nebulized—added to the first-line IV antibiotics).
In agreement with the current trends of personalized medicine,30 we postulated that similar interventions (nebulization of antibiotics) might have divergent effects in different subsets, justifying the formulation of multiple PICO questions with predetermined outcomes. Therefore, the evaluated safety and efficacy predefined outcomes (table 2) were determined as enclosed: (1) adverse events (nephrotoxicity, neurotoxicity and cardiorespiratory complications); (2) emergence of resistance and superinfection; (3) clinical resolution and mortality; and (4) length of intensive care unit stay and mechanical ventilation.
Efficacy outcomes were evaluated according to both the susceptibility pattern of the pathogen and the administration’s strategy. None of the safety outcomes was considered to be possibly influenced by the susceptibility pattern of the pathogen; therefore, this factor was not taken into consideration in their analysis. Evaluation of systemic toxicity was performed according to the administration strategy for better discrimination of their impact. Occurrence of cardiorespiratory complications was considered not to be influenced by the specified administration strategies, therefore this factor was not taken into consideration in its analysis. The effect of nebulized antiinfective agents against viral and fungal infections was also assessed.
A global search strategy was systematically performed in three different databases: MEDLINE database through the PubMed search engine, EMBASE, and the Cochrane Library Database. Search terms are detailed in appendix 1. No restrictions of study design, time, or language were imposed. The first search was performed in June 2014, and it was repeated in March 2015 and July 2016. No other eligible studies were identified by evaluating previous reviews, abstracts from meetings or under suggestion of the panel of experts.
The eligibility criteria, study selection, data collection, and risk of bias assessment are described in detail in the appendix 2. The main characteristics and quality of the studies are summarized in detail in appendix 3. Analysis of all outcomes was performed according to the design of the study, being either randomized controlled trials (RCTs) or observational studies, and their results are also presented accordingly. As the majority of the included studies had a small sample size, a pooled evaluation of all studies was also performed for each outcome, in order to detect a potential presence of clinically significant trends. This approach was considered acceptable due to the lack of large-scale data.
All statistical analyses were performed using Review Manager (RevMan) version 5.3. (Nordic Cochrane Centre, Cochrane Collaboration, Denmark, 2014). The summary statistic measures used for the evaluation of binary outcomes were the risk ratio for RCTs and the odds ratio (OR) for the observational study and pooled evaluation. Risk difference was also used where necessary. The summary statistic measure used for the evaluation of continuous outcomes was the mean difference. All statistical measures were calculated with 95% CI. Random-effects meta-analysis using the Mantel–Haenszel model approach was chosen to obtain pooled study results. The Higgins I2 test was predefined to quantify heterogeneity (I2 ≤ 25% for low, 25% < I2 < 50% for moderate, and I2 ≥ 50% for high). Metaregression was not performed given the low number of studies included in the analysis. Assessment of publication bias using a funnel plot31,32 was planned when considered meaningful (i.e., at least 10 studies available).
A total of 1,435 studies were identified: 898 studies in the MEDLINE database (PubMed), 327 in EMBASE, and 210 in the Cochrane Library Database. After assessment for inclusion, manually adjusting for duplicates, and revision of the articles, 11 studies were finally included in the meta-analysis. The PRISMA flow diagram33 of the studies’ selection is presented in figure 1.
No RCTs or observational studies were found to evaluate the efficacy and/or safety of nebulized antivirals or antifungals for the treatment of respiratory viral and fungal infections in mechanically ventilated patients. Characteristics and results on efficacy and safety of their use are detailed in appendixes A4 and A5.
Six RCTs and five observational studies assessed the efficacy and safety of nebulized antibiotics for the treatment of bacterial infections, involving 826 patients. Five of the studies had a small sample size (fewer than 50 patients). The largest study was an observational study involving 208 patients. Five studies administered aminoglycosides, four administered colistin, and two administered both. Four studies used jet nebulizers, three studies used vibrating-mesh nebulizers, and three studies used different devices indistinctively (jet and ultrasonic nebulizers were used in two studies; the other study used jet and vibrating-mesh nebulizers). One of the studies did not detail the type of nebulizer used, but they did specify having used a device generating optimal-size droplets with diameter between 1 and 5 μm. The risk of bias of the included studies was globally low. The main characteristics of the included studies and the administration strategy to which they belong are described in table 3. A summary of their risk of bias of the included RCT is detailed in figure 2.
Treatment of VAT.
Only two RCT by Palmer et al.34 and Palmer and Smaldone35 involving 85 patients evaluated the efficacy and safety of nebulized antibiotics. Both studies assessed efficacy, but neither of them did so in our predefined terms (both considered reduction of the Clinical Pulmonary Infection Score as a sign of clinical resolution). Meta-analysis of the results showed a greater reduction of the Clinical Pulmonary Infection Score, with a high heterogeneity (mean difference, −3.11 points; 95% CI from −6.18 to −0.04; I2 = 90%), and significantly less emergence of resistant strains in patients receiving nebulized antibiotics (70 patients; risk ratio = 0.18; 95% CI, 0.05 to 0.64; I2 = 0%). The forest plot is shown in figure 3. No significant differences were found in mortality and duration of mechanical ventilation. No significant difference in systemic toxicity (nephrotoxicity, indirectly measured by serum creatinine concentration) was found.
Efficacy of Nebulized Antibiotics for the Treatment of VAP Caused by Resistant Pathogens.
Six studies were included in the analysis of the efficacy of nebulized antibiotics for the treatment of VAP caused by resistant pathogens.
Adjunctive Administration Strategy.
Four studies administered nebulized antibiotics under this strategy; one RCT, Niederman et al.36 (administering nebulized amikacin in patients who could already be receiving IV aminoglycosides), and three observational studies: Kofteridis et al.,37 Doshi et al.,38 and Tumbarello et al.39 Efficacy was assessed by all of the included studies.
No significant difference in clinical resolution was evidenced in the RCT (48 patients; OR, 1.30; 95% CI, 0.22 to 7.55), but it was significantly higher in patients receiving nebulized antibiotics in the meta-analysis of the observational study included (389 patients; OR, 0.51; 95% CI, 0.34 to 0.77; I2 = 0%). Pooled meta-analysis of all studies (437 patients) also showed significantly better clinical resolution rates in patients receiving nebulized antibiotics (OR, 0.53; 95% CI, 0.36 to 0.80; I2 = 0%). The forest plot, stratified according to the study design, is shown in figure 4.
Two observational studies37,38 reported VAP-related mortality, which was found to decrease in patients receiving nebulized antibiotics (181 patients; OR, 0.5; 95% CI, 0.26 to 0.96; I2 = 0%). All-cause mortality was reported by two observational studies37,39 with no difference in the mortality rates (294 patients; OR, 0.68; 95% CI, 0.33 to 1.37; I2 = 46%), with moderate heterogeneity. The forest plot for mortality is shown in figure 5.
A significant decrease in the length of mechanical ventilation support to patients receiving nebulized antibiotics was also evidenced (303 patients; 3.72 days fewer; 95% CI from −5.86 to −1.59 days; I2 = 0%) in the meta-analysis of the two observational studies that assessed it.38,39 The forest plot is shown in figure 6. No other significant differences were found in the rest of the efficacy outcomes assessed (duration of intensive care unit stay and development of superinfection).
Substitution Administration Strategy.
One observational study administered nebulized antibiotics under this strategy: Ghannam et al.,40 involving 32 patients. Higher rates of clinical resolution in patients receiving nebulized antibiotics were reported (OR, 9.53; 95% CI, 1.85 to 49.2). Duration of mechanical ventilation and intensive care unit stay were similar, independently of the use of nebulized agents.
Efficacy of Nebulized Antibiotics for the Treatment of VAP Caused by Susceptible Pathogens.
One RCT, Lu et al.,26 was analyzed. In this trial, both aminoglycosides and ceftazidime were administered either by nebulization or IV, with no other IV antibiotics being administered concomitantly, which was considered to be equivalent to the substitution administration strategy.
No significant differences were apparent for clinical resolution, mortality, duration of mechanical ventilation (median of 11 more days; P = 0.13), intensive care unit stay (median of 9 more days; P = 0.08), and development of superinfection between the patients receiving nebulized antibiotics and the patients receiving systemic therapy. However, per-treatment emergence of resistant strains was considered to be potentially prevented by antibiotics nebulization, as new growth or persistence of infection was caused exclusively by susceptible strains in patients treated with nebulized antibiotics, while 50% of the strains became intermediate or resistant in the group of patients treated systemically.
Safety of Nebulized Antibiotics for the Treatment of VAP.
Evaluation of Systemic Toxicity of Nebulized Antibiotics Administered according to the Adjunctive Strategy.
Two types of systemic toxicity were reported: nephrotoxicity36,37,39,42,43 and neurotoxicity.37,42 No significant difference was evidenced in the occurrence of nephrotoxicity (fig. 7A) or neurotoxicity between patients receiving nebulized antibiotics and patients without intratracheal therapy.
Evaluation of Systemic Toxicity of Nebulized Antibiotics Administered according to the Substitution Strategy.
Two studies reported nephrotoxicity events.40,41 The RCT40 did not show any difference in the risk of nephrotoxicity (risk difference, −0.40; 95% CI, −0.85 to 0.05), but the analysis of the observational study41 did show less occurrence of nephrotoxicity when nebulized antibiotics were administered (risk difference, −0.31; 95% CI, −0.55 to −0.08). The pooled analysis of both (involving 42 patients) also showed significantly less occurrence of nephrotoxicity in patients receiving treatment with nebulized antibiotics (risk difference, −0.33; 95% CI, −0.54 to −0.12; I2 = 0%; fig. 7B).
All studies reporting data on cardiorespiratory complications were analyzed together, according to the study design, but independently of the administration strategy. The meta-analysis of two observational studies37,43 showed no differences in the occurrence of cardiorespiratory adverse events, with no heterogeneity (risk difference, 0.00; 95% CI from −0.04 to 0.04; I2 = 0%). Meta-analysis of the four RCTs included26,36,41,42 showed a 9% increase in incidence of respiratory complications in patients receiving nebulized antibiotics (risk difference, 0.09; 95% CI from −0.01 to 0.18; I2 = 52%), but high heterogeneity between studies was observed (fig. 8). The combined analysis of all studies was similar but with higher heterogeneity (risk difference, 0.04; 95% CI, from −0.02 to 0.11; I2 = 75%). The study that registered more cardiorespiratory complications26 detailed that no bronchospasms were evidenced. Three patients with an initial severe hypoxemia (Pao2/Fio2 ratio of less than 200) suffered a 25% decrease in their Pao2 after the nebulization. Three other complications were related to obstruction of the expiratory filter, detected in two of the cases by an increase in the peak airway pressure. The other patient suffered a secondary cardiac arrest. Finally, one patient had to be excluded in an early phase of the study due to severe hypoxemia secondary to alveolar derecruitment induced by the nebulization.
Our study provides new information on the efficacy and safety of antibiotic nebulization in mechanically ventilated patients. This is important because despite its administration being increasingly common practice worldwide,27 our study demonstrates limited available evidence for its use.
According to our analysis, in terms of efficacy, the administration of nebulized antibiotics might increase the likelihood of clinical resolution (particularly in VAP caused by resistant pathogens), but this is not consistently translated into a significant improvement in mortality or mechanical ventilation duration. Antibiotic nebulization also appears to have a protective effect against the emergence of resistant strains when used for the treatment of VAT or even VAP caused by susceptible pathogens. This contrasts with previous studies, reported before 1985,1–6 probably due to technical limitations in the delivery of the drug and the prolonged administration periods at that time. In terms of safety, our analysis reveals that the risk of nephrotoxicity might be lower when the antibiotics are nebulized instead of administered IV. In July 2016, a new systematic search of the literature was performed, identifying one new RCT by Abdellatif et al.,44 performing colistin nebulization under a strategy equivalent to the substitution strategy. Due to limitations in its design, it did not meet our eligibility criteria for its efficacy evaluation, but we evaluated their data regarding safety (nephrotoxicity), which was consistent with the results of our analysis (fig. 7B), suggesting less occurrence of nephrotoxicity when nebulized antibiotics are administered (for both RCT included: Risk difference, −0.23; 95% CI, −0.37 to −0.10; I2 = 0%; for pooled analysis of all: Risk difference, −0.25; 95% CI, −0.37 to −0.14). Finally, our study also reveals a 9% increase in risk of respiratory complications, especially when they are administered to severely hypoxemic patients, such as severe acute respiratory distress syndrome patients.
The main limitations of our study are its small sample size and the fact that only half of the included studies were RCTs. Due to the small sample size of the studies, no subgroup analyses could be performed, such as a comparison of the efficacy among the three different types of devices used or between the different drugs administered. As the number of RCTs included in the meta-analysis was also very small, a meta-regression could not be performed. Two of the RCTs were not blinded and for the analysis of one of the PICO questions (regarding treatment of VAT), all the included studies were from the same investigator, which introduces a possibility of bias on the results of that specific analysis. However, this is not only a limitation, but an interesting denouement, as it evidences the lack of RCTs in this field.32,45
Another limitation is the fact that all of the included studies in our meta-analysis were published between 2007 and 2015. Clinical studies published before 2014 may have used infratherapeutic doses of the drugs, as the recommended doses of colistin and aminoglycosides have markedly increased the last years based on pharmacokinetics/pharmacodynamics studies.23,46,47 Also, continuous renal replacement therapy might be a confounder, requiring a substantial modification in the administered doses.48,49
It is also likely that some overlap is present between VAT and VAP in the analyzed studies, given the difficulty in their diagnosis.50 Moreover, for the analysis of one PICO question (regarding VAT), the Clinical Pulmonary Infection Score used as an outcome for clinical resolution in the included studies was not a predefined outcome of our analysis. However, as the Pao2/FIO2 is a predictor of mortality and clinical resolution in VAP,51 and as both entities—VAP and VAT—are closely linked and overlapping, we considered that a decrease in Clinical Pulmonary Infection Score in patients treated for VAT was fulfilling the third criteria of our predefined terms of clinical resolution (table 2) and included these studies in the analysis. Another limitation was that some of the outcomes were surrogated, and some of the included studies had heterogeneous populations, with potentially higher risk of mortality and/or morbidities than the average population included (e.g., oncologic patients, delayed targeted antibiotic therapy). Finally, our study protocol was not previously published or registered in a platform like PROSPERO (International Prospective Register of Systematic Reviews, https://www.crd.york.ac.uk/PROSPERO/). However, all clinically relevant questions were identified and defined a priori, using the PICO format, and the initial protocol was never modified. Meta-analyses may result in type I errors owing to an increase of random error when sparse data are analyzed.52
The main strength of our meta-analysis is our strict inclusion of studies. The nebulization devices used in the included studies were restricted to the ones providing sufficiently small particles to reach the lung parenchyma (jet, ultrasonic, and vibrating-mesh nebulizers). Similarly, the population of the included studies was also strictly selected (e.g., only studies involving invasively mechanically ventilated patients were included, studies with colonized-but-not-infected patients were also excluded). Our strict selection limited the study size as a drawback, but also makes our results more robust and generalizable to the study population, in contrast with previous reviews53–57 that had highly heterogeneous results and a potential overestimation of the effects. This is the reason why VAT and VAP were analyzed separately, contrasting with recent systematic review and meta-analysis.53–57 Our study is also the first to analyze different administration strategies (adjunctive and substitution) separately. Finally, our analysis adds value regarding important information on safety aspects such as respiratory or nephrotoxicity adverse events.
In conclusion, our study shows that very limited evidence exists on the use of nebulized antibiotics in mechanically ventilated patients. Improvement in clinical resolution does not translate to improvements in other significant outcomes, which should be enclosed in further studies as predetermined outcomes. Patients with resistant pathogen-related infections might potentially derive greater benefit from nebulized antibiotic therapy. Its use, without the concomitant IV administration of the drug, may reduce nephrotoxicity associated with systemic colistin or aminoglycosides. Administration of nebulized antibiotics seems to be associated with a higher risk of respiratory complications, particularly in severely hypoxemic patients. Therefore, future studies should stratify patients based on the degree of hypoxemia. Our findings evidence that in an era of emerging Gram-negative–resistant organisms, further research with larger RCT including more homogeneous populations, standardized drug delivery, and clinically relevant predetermined outcomes is an urgent and unmet clinical need.
The results of the systematic review and meta-analysis will be translated into recommendations following the GRADE (Grading of Recommendations Assessment, Development and Evaluation) system as part of an Institutional Position Paper convened by the European Society of Clinical Microbiology and Infectious Diseases (ESCMID, Basel, Switzerland) Study Group for Infections in Critically Ill Patients (Basel, Switzerland) under the support of the ESCMID and in collaboration with the Iberoamerican Cochrane Center in Barcelona, Spain. The authors acknowledge Pablo Alonso, M.D., Ph.D., Ivan Solà, M.D., and Sandra Pequeño, M.D., of the Iberoamerican Cochrane Center, for their assessment on methodology. Ivan Solà, M.D., also acted as an independent reviewer for the inclusion of studies, and Sandra Pequeño, M.D., acted as an independent reviewer for the bias risk assessment.
This project has received funding from the European Society of Clinical Microbiology and Infectious Diseases (ESGCIP and EPASG; Basel, Switzerland), Fundació Catalana de Pneumologia (FUCAP; Barcelona, Spain), Centro de Investigación (CIBERES; Madrid, Spain), and European Regional Development Fund (FEDER; European Commission, Brussels, Belgium).
Dr. Rello received research grants and consulting fees from Bayer (Leverkusen, Germany) and Genentech (San Francisco, California). Dr. Roberts received consulting fees from Infectopharm (Heppenheim, Germany). Dr. Chastre received honoraria for lecture or advisory board from Bayer, Pfizer (New York, New York), Basilea (Basel, Switzerland), Astra-Zeneca (London, United Kingdom), Cubist-MSD (Lexington, Massachusetts), MSD (Kenilworth, New Jersey), Kenta-Aridis (San Jose, California), and Medimmune (Gaithersburg, Maryland). Dr. Palmer received research grants from Nektar Therapeutics (San Francisco, California) and consulting fees from Bayer and holds patents for the endobronchial delivery of antibiotics in ventilated patients through the Research Foundation of Stony Brook (Stony Brook, New York) and participated in the 2016 Infectious Diseases Society of America/American Thoracic Society Guidelines Committee for ventilator-associated pneumonia/hospital-acquired pneumonia. Dr. Luyt received honoraria for lecture or advisory board from Bayer, MSD (Kenilworth, New Jersey), ThermoFisher Brahms (Waltham, Massachusetts), and Astellas (Tokyo, Japan). The other authors declare no competing interests.
Appendix 2: Appendix to “Materials and Methods”
Randomized controlled trials, observational studies, and case series evaluating efficacy and/or safety of nebulized antiinfective agents for the treatment of respiratory infections in invasively mechanically ventilated adult patients were eligible if the delivery of the drug was performed with devices generating particles smaller than 5 μm of diameter: jet nebulizers, ultrasonic nebulizers, or vibrating-mesh nebulizers.
Exclusion criteria included studies involving pediatric patients, patients without invasive mechanical ventilation support, patients diagnosed as being colonized but not infected, and patients with particular characteristics such as burn-injured patients or patients receiving support with renal replacement therapies and/or cardiopulmonary support with extracorporeal life support devices such as extracorporeal membrane oxygenation, due to the lack of knowledge on the impact these techniques might have in the technique being evaluated. Studies involving patients with cystic fibrosis or other non-cystic fibrosis were also excluded as they were considered to have particular characteristics deserving a separate evaluation. Studies reporting delivery with other devices, or other practices such as tracheal instillation (either manually or with a pneumatic pump), were rejected as they produce larger particles that may not sufficiently reach the lung parenchyma.
A list of efficacy and safety outcomes to be evaluated (table 2) was independently rated by all authors according to their potential clinical relevance or impact on answering the Population-Intervention-Comparison-Outcome questions. Outcomes were classified as being “nonimportant” (rated 1 to 3), “important” (4 to 6), or “critical” (7 to 9). Only critical outcomes (with a mean score equal to or more than 7) will be evaluated under the Grading of Recommendations Assessment, Development and Evaluation methodology.
Three authors (S.B., G.P., and C.S.-L.) independently assessed all the studies identified in the literature search by screening their titles and abstracts. Disagreements between reviewers were resolved by consensus. In case of disagreement, a fourth independent reviewer (I.S.) determined the eligibility. Authors of articles considered for rejection due to lack of information (e.g., type of device used) were contacted to provide further details.
Data Collection Process
Full texts of the selected studies were obtained. A data sheet, based on the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions,23 was developed for data extraction, which was performed by one author (C.S.-L.) and afterward checked by a second independent reviewer (S.P.). Authors of articles with relevant nonreported or unclear data were contacted to provide further information.
For each included study, the following data were extracted: general information regarding the study design, inclusion/exclusion criteria, etc.; type of intervention performed (including the type of device for delivery); main and secondary outcomes evaluated and adverse events reported.
Risk of Bias Assessment
Risk of bias was assessed for every included study by one author (C.S.-L.) and afterward checked by a second independent reviewer (S.P.) from the Iberoamerican Cochrane Center, Barcelona, Spain. Assessment was performed based on the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions31 : bias regarding selection, performance, detection, attrition, and reporting was assessed for all randomized controlled trials. Bias risk was also assessed for observational studies under the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions,31 evaluating selection and representativity of the cohorts, presence of confounding factors, and adequacy of both the outcomes and their following process.
Appendix 3: Main Characteristics and Quality of the Included Studies
Appendix 4: Study Characteristics and Meta-analysis on Efficacy and Safety of Nebulized Antivirals for the Treatment of Viral Infections
No randomized trials or observational studies were found to evaluate the efficacy and/or safety of nebulized antivirals for the treatment of viral infections; therefore, no evidence can be provided on their use. Only some case series and reports on nebulization of zanamivir were found.58 Zanamivir is not approved for nebulization, and the Food and Drugs Administration alerted in October 2009 (http://www.medscape.com/viewarticle/710336) of the death of a person with influenza who had received its powder for inhalation formulation through a nebulizer. According to the manufacturer, lactose sugar in the formulation increases the risk of obstruction of the mechanical ventilator circuit.
Appendix 5: Study Characteristics and Meta-analysis on Efficacy and Safety of Nebulized Antifungals for the Treatment of Fungal Infections
No randomized controlled trials or observational studies were found to evaluate the efficacy and/or safety of nebulized antifungals for the treatment of fungal infections. Only one case series59 reported their experience on nebulization of Amphotericin B Lipid Complex (Sigma Tau Pharmaceuticals, USA) to 32 immunosuppressed oncology patients as adjunctive therapy to systemic antifungals. Only eight patients were under mechanical ventilation support and had a demonstrated fungal infection. Only one of the patients was reported to have survived. They also reported respiratory complications (modest cough, mild bronchospasm, and transient chest pain) in three patients although it was not specified if those patients were under mechanical ventilation or not. No other evidence was found for the use of nebulized antifungals for the treatment of fungal respiratory infections.